I LARGE-S I RES ELECTRIC P S

Erasmus Mundus Joint Master in Economics and

Management of Network Industries (EMIN)

Master’s Thesis

IMPACT OF LARGE-SCALE INTEGRATION OF RES IN

ELECTRIC POWER SYSTEMS

EXPLORATION OF THE FUTURE EUROPEAN ELECTRICITY MARKET DESIGN

Author: Ainhoa Villar Lejarreta

Supervisor: Remco Verzijlbergh

Co-Supervisor: Germán Morales-España

Madrid, January 2016

UNIVERSIDAD PONTIFICIA COMILLAS

ESCUELA TÉCNICA SUPERIOR DE INGENIERÍA (ICAI)

DELFT UNIVERSITY OF TECHNOLOGY

Page left blank intentionally

Erasmus Mundus Joint Master in Economics and Management of Network Industries (EMIN), 2013-2015

This master thesis is part of the requirements of the Erasmus Mundus Joint Master in

Economics and Management of Network Industries (EMIN), 2013-2015. Delft

University of Technology and Comillas Pontifical University are the two participating

universities. In the end of this Erasmus Mundus EMIN programme, the student will be

granted for both the Master degree of Electric Power Industry from Comillas Pontifical

University, and the Master degree of Engineering Policy Analysis from Delft University

of Technology.


20160111


Exploration of the future European Electricity Market Design

Summary

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta i

SUMMARY

Until recently the European electricity system has been operating as a set of isolated national

markets with divergent regulatory norms. Today, the day-ahead markets between South-

Western Europe and North-Western Europe are fully coupled, enabling the trading of electricity

all the way from Portugal to Finland. Moreover, policy-makers in the power sector are currently

preparing the design of what it will be called the Internal Energy Market. Europe's power sector

aims for an integrated, more competitive, secure and sustainable power system. Meanwhile,

ambitious renewable targets aim for a decarbonization of the electricity sector by 2050. This

will lead to a large deployment of renewable technologies into the system, with almost 50% of

them being the most intermittent, uncertain and unevenly distributed sources in the continent,

wind and solar.

The introduction of the projected large amounts of intermittent sources will impact both the

functioning of the electricity markets and the operation of the transmission grids. Cross-border

congestion profiles are expected to suffer changes, transmission constraints will appear and an

effective congestion management approach will be needed within the framework of the market's

redesign.

Currently, the congestion management mechanism in place in many Member States is based on

a redispatch phase after market clearance, which results in inefficiencies and additional

rebalancing costs. This congestion management approach falls within the zonal market design.

A consistent integration of electricity markets across Europe enabling the access of large

capacities of renewable generation would have the potential to maximize overall welfare to all

agents. Generators would enter a more competitive market with a lower risk, consumers would

benefit from lower electricity prices and transmission system operators would benefit from

reduced operation costs of balancing and reserve.

Locational marginal pricing, also known as nodal market design, would be able to provide an

integrated approach of national and international congestion management, a joint allocation of

international transmission rights, the integration of congestion management with day-ahead,

intraday and balancing markets and finally a transparent approach to facilitate secure and

effective cooperation and information exchange among European system operators. However, a

committed high-level support on a European level would be required for its further

implementation.

The objective of this thesis is to gain a better understanding of the European power market in

presence of large amounts of renewable energies and under different power market designs, and

find out by how much can the market design and its features affect the provision of cost-

efficient electricity, this is the system's variable generation costs of electricity.

In order to study the influence of the power market design in the integration of large amounts of

renewable energies, an optimization model is used. The model solves a weekly unit


Summary

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta ii

commitment problem and the transmission constrained economic dispatch for the day-ahead

market of a conceptual network of Europe and from a centralized decision making point of

view. The model uses a mixed-integer linear programming (MILP) formulation of the unit

commitment and minimizes total variable generation costs of the system. The two market

designs studied, nodal market and zonal market, are modeled according to how congestion is

managed in each case.

The aim is to analyze the impact of the power market design on the power system's variables:

total variable generation costs, RES curtailment, energy production by technology, non-served

energy and hourly electricity prices while subjected to several scenarios of increasing degree of

RES integration. For this, different degrees of future RES scenarios based on ENTSO-E market

studies are used.

The research shows that in a high renewable scenario the total variable generation costs of the

power system when it operates under a zonal power market are around 0,32% higher than under

a nodal market. These potential savings under a nodal market could even be larger especially if

the large expected projections of renewable sources of generation finally materialize and

provided that the required network capacities are delivered effectively on time.

Moreover, the degree of curtailment in both nodal and zonal markets rise notably in a high

renewable scenario compared to the current situation, up to a weekly curtailment of 7,83% and

8,01%, respectively. Such notable amounts of curtailment could be due to the insufficient

development of the transmission network assumed which leads to the incapability of supplying

cheap renewable energy across wide regions and instead having to commit or reschedule local

and more expensive technologies.

On the other hand, costs of unserved energy represent 2,95% and 3,08% of the total system's

variable generation costs in the nodal and zonal market, respectively. It is again highlighted the

importance of a timely delivery of the network infrastructure investments to gradually integrate

the large deployment of renewables in the system.

Overall, a nodal market in Europe could have the potential of improving efficiency in the

system by reducing variable generation costs by 0,32% compared to a zonal market. Benefits

could increase even more if an adequate network expansion plan that takes into account the

growth of renewable energies would deliver its investments on a timely manner.


Preface

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta iii

PREFACE

This research thesis is the end point of the two-year Erasmus Mundus Master Programme

Economics and Management of Network Industries (EMIN), carried out at the faculty of

Technology, Policy and Management of the Delft University of Technology (Delft) and at the

ICAI Engineering School of Comillas University (Madrid).

The research project presented below is aimed at everyone interested in electric power systems,

in the integration of renewable energies and the design of power markets.

Upon completion of this thesis, I would like to thank the European Commission for giving me

the opportunity to pursue the EMIN programme and allowing me to combine my studies in TU

Delft and Comillas University. I would also like to appreciate professors in both universities for

their valuable teachings in the fields of energy and policy analysis. In particular, I would like to

thank my first supervisor Remco for his constant support and guidance and Germán for his

valuable input in the preparation of this thesis, as well as Bert Enserink and Pauline Herder for

their constructive feedback.

I also want to thank my friends Hugo, Shujie, Roxanne, Kevin, Donna, Hector, Hanif, Lien,

Djong, Viet, Maria Luisa, Mariam and Tariq for their help and shared happiness during this

time. Last but not least, I want to thank all my family, especially my dear parents for their

continuous support and help at all times.

Ainhoa Villar Lejarreta, December 2015.


Preface

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta iv



Table of Contents

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta v

TABLE OF CONTENTS

Summary………………………………………………………………………………………....i

Preface…………………………………………………………………………………………...ii

Table of Contents………………………………………………………………………………..v

List of Figures………………………………………………………………………………….vii

List of Tables……………………………………………………………………………………ix

Nomenclature…………………………………………………………………………………...xi

1. Introduction………………………………………………………….....……………………3

1.1. Europe's Future Power System. Problem in context…………………………………….3

1.2. Impact of large-scale integration of renewable generation………………………...........6

1.2.1. Effects on the day-ahead, intraday & balancing electricity markets…………….6

1.2.2. Effects on congestion management……………………………………………..10

1.3. Power market designs and congestion management…………………………………...11

1.4. Problem definition……………………………………………………………………...12

1.5. Research objectives…………………………………………………………………….13

1.6. Thesis outline and structure…………………………………………………………….14

2. Theory and Model Description……………………………………………………………19

2.1. Theory of Unit-Commitment…………………………………………………………...19

2.2. Theory of Optimal DC Power Flow……………………………………………………21

2.3. Locational Marginal Prices……………………………………………………………..25

2.4. Modeling approach……………………………………………………………………..26

2.4.1. Modeling Objectives………………………………………………………....…27

2.4.2. Modeling Methodology: The Day-Ahead Market…………………………...…27

2.4.3. Modeling Methodology: Implementation of Power Market Designs………..…30

2.5. The Optimization Problem. Model Formulation…………………………………….…32

2.5.1. Objective Function………………………………………………………...……34

2.5.2. Constraints………………………………………………………………………36

2.5.2.1. Power system balance requirement……………………………………36

2.5.2.2. Production of committed units………………………………………...36

2.5.2.3. Logical commitment and minimum up and downtime constraints……37

2.5.2.4. Generation capacity limits………………………………………….....37

2.5.2.5. Upward and downward system operating reserves……………………38

2.5.2.6. Ramping limits………………………………………………………...38

2.5.2.7. Network constraints…………………………………...………………39

3. Simulation Setup……………….....………………………………………………………..43

3.1. Scenario definition……………………………………………………………...………43

3.1.1. Reference scenario……………………………………………………………...44

3.1.2. Scenario EU 2020………………………………………………………………45


Table of Contents

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta vi

3.1.3. The Visions for 2030…………………………………………………………....46

3.2. Model Data………………………………………………………………………….….50

3.2.1. Demand………………………………………………………………………....50

3.2.2. Generator properties………………………………………………………….....51

3.2.2.1. Generator costs…………………………………………….………….51

3.2.2.2. Technical and environmental characteristics……………………….....53

3.2.2.3. Initial conditions………………………………………………….…..55

3.2.3. Renewable generation……………………………………………………..……55

3.2.4. Network parameters…………………………………………………………….57

3.2.4.1. Transmission line capacities………………………………………..…58

3.2.4.2. Inductive reactances…………………………………………………..58

4. Simulation Results………………………………………………………………………....61

4.1. Power system generation costs……………………………………………..………….61

4.2. Further analysis of RES scenarios and power market designs……………..………….65

4.2.1. Curtailment of renewable energy…………………………………...…………..65

4.2.2. System generation dispatch profiles………………………………….…………68

4.2.3. Non-served energy……………………………………………………………....72

4.2.4. Effects of system outputs on the generation costs………………...…………….73

4.2.5. Electricity prices and congestion costs…………………………….…………....75

4.2.6. Sensitivity analysis of model parameters..............................................................78

4.2.6.1. CO2 emission cost rate...........................................................................78

4.2.6.2. Intra-zonal network capacities...............................................................82

4.2.6.2.1. Double intra-zonal network capacities......................................82

4.2.6.2.2. Half intra-zonal network capacities..........................................85

5. Conclusions and recommendations……………………………………………………….91

5.1. Conclusions of the research…………………………………………………………...91

5.2. Reflections on the modeling process and results...........................................................95

5.3. Recommendations……………………………………………………………….….…96

5.3.1. Guidelines for future work…...…………………………………………….……96

5.3.2. Reflection and policy implications in the European framework………………..97

Bibliography…………………………………………………………………………………..101

Appendices…………………………….……………………………………………………...105

Appendix A GAMS model code……......………………….......………………………...107

Appendix B Generation mix of the system, per country and per scenario........……...119

Appendix C Demand parameter and assumptions.....……………………………....…125

Appendix D Generator properties, fuel costs and network parameters……………...127


List of Figures

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta vii

LIST OF FIGURES

Figure 1.1. Price convergence in Central-Western Europe………………………………………7

Figure 1.2. Day-ahead prices in France and power transmissions from Germany to France…….8

Figure 1.3. Cross-border arrangements in intraday and balancing markets in Europe in 2009….9

Figure 2.1. Impedance of line ij…………………………………………………………………22

Figure 2.2. Conceptual European network modeled in the day-ahead market………………….28

Figure 2.3. Schematic representation of the modeling process of the power market designs..... 31

Figure 3.1. General overview of the generation mix across all scenarios………………………44

Figure 3.2 . Breakdown of generation mix per country in Scenario B year 2014………………45

Figure 3.3 . Breakdown of generation mix per country in Scenario EU year 2020…………….46

Figure 3.4. Political and economic frameworks of the four Visions……………………………48

Figure 3.5. Generation and load frameworks of the four Visions………………………………48

Figure 3.6 . Breakdown of generation mix per country in Vision 1 year 2030…………………49

Figure 3.7 . Breakdown of generation mix per country in Vision 4 year 2030…………………49

Figure 3.8. Linear approximation of input-output characteristic of generator g………………..52

Figure 3.9. Global irradiation (kWh/m2) (above) and solar electricity (kWh/kWp) (below) in

European countries……………………………………………………………………………..56

Figure 3.10. Annual full-load hours for onshore wind energy in the European Union…………57

Figure 4.1. Generation costs in Reference scenario B (2014) and Vision 4 (2030)…………….64

Figure 4.2. Distribution of the hourly RES curtailment of the power system in Reference

scenario (2014) and Vision 4 (2030)…………………………………………………………....66

Figure 4.3. Hourly wind and solar energy curtailment in the power system in Vision 4

(2030)…………………………………………………………………………………………...68

Figure 4.4. Generation dispatch profiles in Reference scenario (2014) and Vision 4 (2030) for

the three power market designs…………………………………………………………………70

Figure 4.5. Average Locational Marginal Prices in Reference scenario (2014) and Vision 4

(2030) for the nodal and zonal markets…………………………………………………………76


List of Figures

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta viii

Figure 4.6. Difference in Locational Marginal Prices of nodal and zonal markets in Reference

scenario (2014) and Vision 4 (2030)……………………………………………………………78

Figure 4.7. Generation dispatch profiles in Vision 4 (2030) for the nodal and zonal power

market designs for the base case and low CO2 emission cost rate……………………………..80

Figure 4.8. Average Locational Marginal Prices in Vision 4 (2030) for the nodal and zonal

markets for the base case and low CO2 emission cost rate……………………………………..81

Figure 4.9. Difference in average Locational Marginal Prices between nodal and zonal markets

in Vision 4 (2030) for the base case and low CO2 emission cost rate………………………….82


markets for the base case and double intra-zonal capacities……………………………………84


in Vision 4 (2030) for the base case and double intra-zonal capacities………………………..85


markets for the base case and half intra-zonal capacities……………………………………...87


in Vision 4 (2030) for the base case and half intra-zonal capacities………………………..…88


List of Tables

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta ix

LIST OF TABLES

Table 4.1. Breakdown of generation, non-served energy and redispatch costs for Scenario B

(2014) and Vision 4 (2030) for the three proposed power market designs.………………….....62

Table 4.2. Comparison of total variable generation costs between Reference scenario (2014) and

Vision 4 (2030) for the three proposed power market designs in all their phases……………...63

Table 4.3. Renewable production, total production and curtailment of renewable energy in

Reference scenario (2014) and Vision 4 (2030) for the three power market designs…………..67

Table 4.4. Breakdown of energy production by technology type in Reference scenario (2014)

and Vision 4 (2030) for the three power market designs……………………….………………71


and Vision 4 (2030) for zonal power market in its two phases…………………………………72

Table 4.6. Comparison of non-served energy costs between Reference scenario (2014) and

Vision 4 (2030) for the three proposed power market designs in all their phases……………...73

Table 4.7. Impact of system outputs on variable generation costs……………………………..74

Table 4.8. Breakdown of generation, non-served energy and redispatch costs in Vision 4 (2030)

for the nodal and zonal power markets and for the base case and low CO2 emission cost rate..79


for the nodal and zonal power markets and for the base case and double intra-zonal network

capacities......................................................................................................................................83

Table 4.10. Breakdown of generation, non-served energy and redispatch costs in Vision 4

(2030) for the nodal and zonal power markets and for the base case and half intra-zonal network

capacities……………………………………………………………………….……………….86


List of Tables

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta x



NOMENCLATURE

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta xi

NOMENCLATURE

A. Indexes and Sets

Generating units, from 1 to

Generating units in with minimum uptime equal to one hour

Location of generating unit in bus

Hourly periods, from 1 to hours

Network buses, from 1 to ( and are used indistinctly)

Subset of , network buses connected to at least one AC transmission line

Subset of , network buses connected only to HVDC transmission lines

Power line circuits, from 1 to

Subset of , network buses and connected by an AC transmission line and circuit ID c

HVDC lines, from 1 to

HVDC line from bus (to bus )

B. Parameters

1) Model parameters related to hourly periods:

Instantaneous demand in hour and in node [GW]

Maximum intermittent generation from wind and solar in hour and in node [GW]

2) Model parameters related to the generating units:

Maximum power output of generating unit g [GW]

Minimum power output of generating unit g [GW]


NOMENCLATURE

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta xii

Initial power output of generating unit g [GW]

Initial power output of generating unit g above [GW]

Initial time status of generating unit g, defined as the time the unit has been online [hours]

Initial commitment status of generating unit g, which is equal to 1 if the unit is online and 0 if it

is offline

Ramp-up capability of generating unit g [GW/hour]

Ramp-down capability of generating unit g [GW/hour]

Minimum time online of generating unit g [hours]

Minimum time offline of generating unit g [hours]

Fuel cost of generating unit g [M€/GWh]

Variable operation & maintenance cost of generating unit g [M€/GWh]

Marginal cost of generating unit g [M€/GWh]

Fixed fuel consumption cost of generating unit g [M€/hour]

Startup cost of generating unit g [M€/GW]

Shutdown cost of generating unit g [M€/GW]

Emission rate of generating unit g [tCO2/GWh]

Cost of CO2 emissions [M€/tCO2]

Cost of non-served energy [M€/GWh]

3) Model parameters related to the network:

Inductive reactance X of line and circuit ID [p.u.]

Slack Bus = 2

Susceptance B of line

Impedance Z of line . Inverse of B: Z = B-1


NOMENCLATURE

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta xiii

Power transfer distribution factor for AC transmission line , circuit ID and node

Maximum power capacity of AC transmission line and circuit ID [GW]

Incidence matrix: Incidence of DC line and node

Maximum power capacity of HVDC line [GW]

Locational Marginal Price (nodal price) in node in hour [M€/GWh]

Dual variable of system balance constraint for every hour [M€/GWh]

Dual variable of the network constraint of the AC transmission line for every hour

[M€/GWh]

Dual variable of the power balance constraint in nodes with only HVDC lines for every

hour [M€/GWh]

C. Variables

1) Free and continuous variables:

Total variable generation costs [M€]. Objective function.

Power flow through AC line and circuit ID in hour [GW]

Power flow through HVDC line in hour [GW]

2) Positive and continuous variables:

Power output of generating unit in hour [GW]

Power output of generating unit above the minimum output in hour [GW]

Operating upward reserve of generating unit in hour [GW]

Operating downward reserve of generating unit in hour [GW]

Intermittent generation from wind and solar in node in hour [GW]

Non-served energy in node in hour [GW]


NOMENCLATURE

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta xiv

3) Binary variables:

Commitment status of generating unit in hour , which is equal to 1 if the unit is online

and 0 if it is offline

Startup status of generating unit in hour , which is equal to 1 if the unit starts up and 0

otherwise

Shutdown status of generating unit in hour , which is equal to 1 if the unit shuts down

and 0 otherwise

Master Thesis MSc Economics and Management of Network Industries (EMIN) MSc Engineering and Policy Analysis (EPA) MSc Electric Power Industry (MEPI) – Ainhoa Villar Lejarreta Page 1

CHAPTER 1

INTRODUCTION




Chapter 1 - Introduction


1. Introduction

Until recently the European electricity system has been operating as a set of isolated national

markets with divergent regulatory norms. The trend now is that these markets are unified into a

single more efficient market. This requires solving a number of technical and regulatory

problems.

Additionally, nowadays there is a growing concern about the environment in all different kind

of fields. International institutions, governments and industries are gradually adopting more

sustainable norms and practices in order to exploit resources at minimum cost. This trend can

also be seen in the growth of renewable energies in the power system. They too present issues in

regard to their integration into the system.

Chapter 1 exposes the path followed by the power systems in Europe, makes an appraisal of the

technical and regulatory problems to be solved in the near future and identifies existing and

alternative power market designs that would contribute to a more economically efficient and

transparent European electricity market.

1.1. Europe's Future Power System. Problem in context.

Nowadays, Europe's Power System is immersed in an energy transition period. Policy-makers in

the power sector are currently envisioning the future power system of what it is lately being

called an Energy Union for Europe. Consultations to experts and stakeholders, and legislative

proposals are being put forward for what it will be the biggest redesign of Europe's electricity

market.

Short after the liberalization process of national energy markets around the 1980s started, driven

largely by economic reasons, the European Union already engaged in a broader approach to the

restructuring of the power markets in the Member States. Setting aside the economic national

goals of every individual country in the region, the European Union focused on the strategic and

political concerns involving the power sector. The high dependency of the Union on external

sources of oil and gas prompted the following three pillars which are still the basis of today's

energy needs: secure an increasing supply of energy from domestic and foreign sources, develop

a more competitive European energy market and support environmental protection and the

development of renewable energy sources (Barroso, 2006). With these goals in mind and

following several European Directives in the 1990s that favoured the liberalization of markets,

the grounds for the creation of a single European energy market were established.

However, the restructuring process of the different national markets towards a more competitive

and efficient one turned out to take longer than expected for several reasons. (Pollitt, 2009)

argues that the privatization of state-owned electricity assets, the opening of the market to

competition, the vertical unbundling of transmission and distribution activities from the




generation and retailing services and the introduction of an independent regulator are all four

stages that should be completed in advance for a successful implementation of a market-based

reform. Nevertheless, a continuous process of interactions between market players and

regulatory authorities in all Member States took place revealing the complexity to reach the

same level of development of all stages in the different countries. Technical, economical and

political barriers towards competitive markets explain the delays and the difficulties in the

process of the reform (Karan & Kazdagli, 2011).

Technical barriers are related to the characteristics of the commodity of electricity. Electricity

depends on a physical grid that reduces liquidity and adds complexity to the operation of

markets. Close coordination is required between the operators of different networks for energy

trade to happen and is well-known that every power system has certain market arrangements

that differ from one another, inevitably running into numerous obstacles.

From the economical and political point of view, national governments are reluctant to give up

complete control over national energy markets due to the importance of security of supply and

the energy sector on the economic development of the rest of the industries. This situation

avoids effective competition from taking place and hence inefficiencies appear, such as the need

to introduce state subsidies to keep an industry functioning (Karan & Kazdagli, 2011).

Moreover, the generation mix in every country is extremely different. Some countries might

rely more on a specific technology than another and this, together with EU mechanisms

implemented supporting certain technologies over others, might emphasize price differences

among regions, certainly giving way to uneven distributions of the gains and, thus, rising

uneasiness in governments to further integrate in a common electricity market.

Despite these barriers, considerable progress has been made thanks to the efforts of stakeholders

and the European Commission, the main driving force and main policy maker in this energy

reform. Relevant milestones have been achieved. Today, the full price coupling of the day-

ahead markets between South-Western Europe and North-Western Europe is in place enabling

the trading of electricity all the way from Portugal to Finland. Similarly, on-going progress is

taking place in Eastern Europe. Moreover, a centralised information platform has been made

available by the European Network of Transmission System Operators for Electricity (ENTSO-

E), where data about electricity markets is published, aiming for greater transparency among all

market players.

Complementarily to the market coupling initiatives, an updated investment plan for network

expansion was released by the ENTSO-E in its Ten-Year Network Development Plan (TYNDP)

2014, in which a total investment of 150 Billion Euros is predicted until the year 2030. In this

regard, only two thirds of the infrastructure projects are being delivered on a timely manner.

Transmission System Operators (TSOs) are facing challenges due to permit procedures or lack

of public acceptance. However, the ENTSO-E emphasizes in its plan the benefits that greater

interconnections would bring to all market players by achieving greater convergence in

electricity prices between regions as well as a significant reduction in these (ENTSO-E, 2014).




Again, committed political support is of utmost importance to achieve a proper European

integrated market.

While in the past years focus in the energy sector shifted from liberalization to the integration of

the national markets, more specifically to the coupling of the day-ahead energy markets, the

priority for the future years is set on decarbonisation policies, security of supply and the

integration of markets closer to real-time like intra-day and balancing markets.

The European Union's renewable energy directive from 2009 is boosting investments in

renewable technologies in such amounts that has caused struggles to conventional generation,

grid operators and regulators to effectively respond to ever-increasing uncertainty levels in the

power system. Given the ambitious environmental energy targets set for 2050, for which

European power systems will have to reduce its contribution to greenhouse gas (GHG)

emissions to zero, still large number of investments in renewable sources are expected in the

whole of Europe. Consequently, an adequate future design of the power market would be one

that fosters more operational flexibility to allow TSOs to efficiently manage the increased

uncertainty in the system.

On the other hand, the coupling of the regional day-ahead markets brought improvements over

market results. Transmission capacity is, since then, used more efficiently and the flow of traded

energy is in accordance to the price differences between regions (Borggrefe & Neuhoff, 2011).

Market coupling favours convergence of prices, which boosts competition and therefore better

quality services and ultimately better prices for consumers. Moreover, (Meeus, et al., 2005)

highlighted that a further stage of market integration is necessary and a cross-border balancing

and intraday markets should be set up. In fact, according to the European Commission, short-

term cross-border markets should be at the core of the European power market redesign. These

markets are able to best capture the value of operational flexibility sought for large-scale

integration of renewables due to their closeness to real-time operation.

Finally, greater degree of liberalization of energy markets, more cross-border interconnections

and larger amounts of renewable technologies in the system require changes to the way the

power system is operated. With a top-down framework, the foundations on which the internal

electricity market in Europe is being built are laid through the development of the network

codes. Elaborated by the European Commission, the ENTSO-E and the Agency for the

Cooperation of Energy Regulators (ACER), these codes provide a complete set of rules on

different areas of the electricity market that will need to be implemented and complied with

across Europe. They will become binding technical EU regulations. When this happens, Europe

will then be one step closer to the realisation of the Internal Energy Market (IEM), and hence, of

its energy goals.




1.2. Impact of large-scale integration of Renewable Generation

The energy target set for 2020 to reach at least a 20% share of energy coming from renewable

sources and the initiative of decarbonising the electricity sector by 2050 will lead to a

significant increase in green energy technologies in the power system. Almost 50% of this

installed capacity will be represented by wind and solar power plants which are, of all

renewable energy sources, the most intermittent, unpredictable and unevenly distributed sources

around the continent (Eurelectric, 2010). The large-scale introduction of these types of

renewable sources will impact both the functioning of the electricity markets and the operation

of the transmission and distribution grids.

The following sections go further into the effects of the expected high integration of renewable

energies on the coupled day-ahead, and current intraday and balancing markets in Europe and

on the system operation and congestion management in the network.

1.2.1. Effects on the day ahead, intraday & balancing electricity markets

In most Member States there are three distinct types of electricity markets currently in place:

day-ahead, intraday and balancing markets. Day-ahead and intraday markets are energy markets

in which power is traded on different timeframes, as their names accurately suggest, while

balancing markets have been traditionally used by system operators to provide reserves and

response capacity to balance the system when unplanned events take place, as in the case of

plant outages, load prediction errors or wind uncertainty.

After the regional coupling of the North-Western and South-Western day-ahead markets in May

2014, a greater harmonization of the wholesale prices is expected in the whole region. Such

convergence of prices was also the initial result after the market coupling in the Central-Western

(CWE) region took place in November 2010 (Figure 1.1).

The current market coupling mechanism uses a common price formation that allows to optimize

the allocation of cross-border interconnection capacities for power exchange. The allocation of

this transmission capacity is done through implicit auctioning. This means that the available

transmission capacity is auctioned jointly with the auctions of electricity in the spot markets,

accomplishing in this way the integration of the different national electricity markets. The

optimized use of the interconnections between countries in theory yields two results: (i) a

general decrease in spot prices and (ii) a general decrease in the differences of spot prices

between regions provided that the required transmission capacities are in place.




Figure 1.1. Price convergence in Central-Western Europe. Source: (Böckers, et al., 2013).

Firstly, prices should decrease because, given the principle of the merit order dispatch, cheaper

power from renewables, for example, can be delivered to other regions that would not have been

possible otherwise. It is worth mentioning that the impact of the renewable targets on the day-

ahead market cannot be decoupled from an increasing coupling of the regional day-ahead

markets since they are part of the same strategy (Eurelectric, 2010). To illustrate the reader,

Figure 1.2 presents the power exported from Germany to France and the day-ahead prices in

France after the market coupling took place in 2010. In the graph, when German imports

increase, French day-ahead prices fall. Knowing that Germany has a higher renewables capacity

share compared to France, it is likely that this price drop is due to cheaper power imports than

its nuclear power, i.e. renewable sources (Doan, 2012).

Secondly, if there is sufficient cross-border capacity and the day-ahead prices in France fall due

to German imports, then spot prices in Germany will increase because there is more power

produced with respect to the situation where there are no imports to France. In this way, a price

equilibrium is reached and price differences are reduced. However, in practice prices are yet not

completely harmonized and price divergence is still significant like in the year 2012 in Figure

1.1. This is due to congestions in the network caused by still insufficient transmission capacity

in that region, unexpected failure of generators, sudden increase in demand or too much

renewable generation in one country.




Figure 1.2. Day-ahead prices in France and power transmissions from Germany to France.

(Source: RTE, Powernext)

In fact, renewable sources in the system lower day-ahead prices on average. However, due to

the stochastic behaviour of weather and therefore the intermittent nature of renewables,

uncertainty levels increase and sudden price spikes can appear when there is no wind or sun

available. As a result, price volatility increases. Moreover, higher penetration of renewables will

increase price volatility even more, influencing the future price hedging strategies and the

behaviour of energy traders in the derivatives market. Furthermore, TSOs will have the difficult

task to decide in real time the most appropriate capacity to dispatch when suddenly the wind

stops or the sun is not available, resulting in higher costs and therefore higher spot prices.

Regarding intraday and balancing markets, several countries in Europe have them in place to

allow for readjustments of the dispatch during the day. However, the market setups differ

significantly from one country to another and are still far from competitive and efficient

harmonized European markets (Figure 1.3) (Borggrefe & Neuhoff, 2011).

The main issue with their current design in most countries is the inability to optimize between

the balancing and the day-ahead markets. Energy suppliers have to send their commitment

offers to either the day-ahead or intraday markets, or to the balancing market. In many cases it is

not possible to change this commitment when we are closer to real-time. Therefore, while

already committed power plants in the day-ahead market would be able to provide upward

balancing power when they are asked to reduce production due to an increase in wind feed-in,

they are unable to do so. Instead, more expensive technologies have to provide the balancing

services.

Moreover, the increased uncertainty in day-ahead markets due to the growing penetration of

renewables has encouraged an increase in demand for reserve capacities in these markets to




ensure security of supply. This results in larger amounts of startups and part-load costs, thereby

increasing generation and operational costs of the system (EWIS, 2010). This trend is expected

to continue with an increasing share of renewables, and despite the improvements in wind

forecasting this will not enable a full use of the system's flexibility if no changes are made in the

design of these markets.

Figure 1.3. Cross-border arrangements in intraday and balancing markets in Europe in 2009.

Source: EWI, 2010.

The initiative of regional coupling between the European day-ahead markets now extends to

applying such coupling mechanism to both intraday and balancing markets. Until now this had

low priority in many countries. (Smeers, 2008) argues that day-ahead, intraday and real-time are

different steps of a single trading process and therefore they require a single trading platform,

instead of the current three different market schemes. The future market design should allow

TSOs to efficiently manage the increased uncertainty in the system and generators should help

by allowing them to offer a joint bid for energy production and provision of balancing services.

The main beneficiaries of the reviewed market design would be European consumers who

benefit from improved security of supply and lower system costs, which results in lower

electricity prices. System operators also benefit from more transparent operational procedures

and reduced costs. However, the improvement and harmonization of the electricity markets can

be seen as a threat to dominant generation companies who might see their large benefits shrink

due to an easier entry of competing generators in the intraday and balancing markets (Borggrefe

& Neuhoff, 2011).




1.2.2. Effects on congestion management

Congestion management has become an operational challenge in more and more liberalized

power systems due to the increasing number of bilateral contracts for electricity trade.

Transmission congestion is the operating condition in which there is insufficient transmission

capacity to deliver all the traded energy simultaneously and therefore certain lines in the

network may become overloaded. TSOs, as operators of the system, are in charge of alleviating

these situations and restoring a secure state of the system.

The further deployment of renewable energies in the following years is expected to take place

mainly in offshore sites, away from load centers, thus new transmission lines will be required.

Besides, distribution grids will also need new investments due to a large increase of distributed

generation.

In the case that the large introduction of renewables is not accompanied by a timely delivery of

the required transmission investments, as already mentioned in section 1.1, transmission

constraints would appear affecting system operation. Congestion already exists in European

cross-border interconnections, but the expected renewable penetration will change this cross-

border congestion profiles to a greater extent (Neuhoff, et al., 2011).

Currently, the most typical way TSOs solve congestions in the European network is by

redispatching generation, curtailing demand or a combination of both. The main concerns of the

system redispatch solution are its high costs, which are usually and ultimately paid by the

consumers, and its susceptibility to high levels of market power. Also, even if generators are

allocated with the redispatch costs, incentives to locate outside congested areas would be

relatively low, resulting in a long-term economically inefficient solution to congestion

(Hakvoort, et al., 2009).

Therefore, an adequate congestion management mechanism should be able to promote the

efficient use of the existing transmission capacity while maintaining a reliable and secure power

system, and guaranteeing maximum transparency to public and private investment agents.

According to (Neuhoff, et al., 2011) an effective congestion management scheme for Europe

would have to integrate the following criteria:

An integrated approach of national and international congestion management

Joint allocation of international transmission rights

Integration of congestion management with day-ahead energy markets

Integration of congestion management with intraday and balancing markets

A transparent approach to facilitate secure and effective cooperation and information

exchange among European system operators




1.3. Power market designs and congestion management

In previous sections it has been illustrated the improvements, the future challenges and the

importance of a closely integrated operation of the power system to achieve a secure,

competitive and sustainable European electricity system. The network codes, developed by the

European Commission, the ENTSO-E and ACER, set the rules and procedures to trade

electricity across Europe based on some framework guidelines and hence define the new power

market design. The power market design proposed by European stakeholders is, as briefly

mentioned in section 1.1, based on implicit auctioning and flow-based capacity allocation.

Currently in Europe, the market design is based on a multi-region day-ahead market coupling

mechanism, or also known as zonal market. The zonal market approach assumes one electricity

price per market zone. In continental Europe every market zone is limited by national

boundaries and in the Nordic region every country is divided into several market zones. By

doing this, intra-zonal congestion is not accounted for through the market. Instead, only cross-

border interconnections are considered and the TSOs manage congestion inside the zones after

market clearance, through a redispatch process, incurring in operational inefficiencies as already

explained. Now, the allocation method of cross-border transmission capacity is not

straightforward mainly because commercial energy flows do not correspond with the actual

energy flows in the network. The latter follow Kirchhoff's circuit laws. There are different

capacity allocation methods used in worldwide electricity markets, namely the Available

Transfer Capacity (ATC) method, the Flow-Based Market Coupling (FBMC) method, both for

the zonal market approach, and the nodal market.

Since May 2015, the FBMC model is currently implemented in the day-ahead market of Central

Western Europe (Belgium, the Netherlands, France, Germany and Austria). In the rest of

continental Europe transmission capacity is still allocated through the ATC methodology. The

difference between these two cross-border capacity allocation methods is at what point in time

the allocation of capacity is done. In the FBMC method the allocation is partly done with the

clearing of the market while in ATC it is done ex-ante. As a result, given the heuristic approach

with which the ATC is calculated and the independence of each cross-border link in the

calculation, the ATC value is conservative to prevent line overloadings, thus reducing the

capacity available to the market (Van den Bergh, et al., 2015). On the other hand, because the

FBMC considers the critical lines in the grid simultaneously with the market, the allowed

commercial capacity between zones is no longer independent from one another and the final

value is less conservative. FBMC method results in a more efficient design for capacity

allocation than the previous ATC method. (Smeers, 2008) points out that the FBMC

methodology is a reliable allocation method and applicable, from a computational perspective,

for all markets from day-ahead to real-time. Despite the improvement in efficiency, the complex

calculations that entail the FBMC method questions its transparency towards market players.

Even though the third allocation method, the nodal market, is not viewed among the targets of

the Internal Electricity Market model, according to (Neuhoff, et al., 2011) it offers an even more




efficient and transparent approach to congestion management. In many power systems in the US

(ERCOT, CAISO, ISO-NE, NYISO, and PJM markets), this option was chosen over zonal

market due to the changing nature of the congested lines in their highly-meshed networks. This

method takes into account all physical transmission constraints in the market clearing, therefore

every node in the network represents one market zone, i.e. one price. Nodal market scheme is

also known as Locational Marginal Pricing (LMP). The price in every node reflects the

locational value of energy, it includes the cost of supplying the energy as well as the cost of

delivering it. Price differences in nodes show the costs of transmission at the same time it gives

proper allocation signals to market players on where to site required generation, transmission

and load. Moreover, since there is no need for a redispatch phase after market clearance, gaming

opportunities and abuse of market power would be significantly reduced. Equally important, the

link between the day-ahead market and the intraday and balancing markets would be improved

(Borggrefe & Neuhoff, 2011).

(Neuhoff, et al., 2011) describe the potential benefits that this market design would add to the

European context given the change in the generation portfolio and the need for a more efficient

use of transmission capacity. Regarding this, it could be useful to rethink in a new zone

configuration for Europe. They argue that the costs of changing the power market design and

establishing new trading arrangements can be high but in case of implementing the change the

sooner it is done the cheaper it would be. Nevertheless, some incumbent generation companies

might be opposed to the change because of rent re-allocation issues. Consumers would be the

great beneficiaries due to lower system costs that would translate into lower electricity prices.

Nevertheless, more initiative at a European level and a committed high-level support among

Member States would be essential for further steps.

1.4. Problem definition

Renewable energy technologies are becoming an increasingly important source of electricity

production in Europe's power system, which is making an important effort to move towards a

decarbonized, secure and more affordable system.

Alongside the benefits brought by the introduction of more green technologies, the uncertain

nature of these sources impact the functioning of the markets and the operation of the system in

all time scopes. In relation to this research, the daily variability of wind and sun influences the

unit scheduling in the day-ahead markets, in which flexible and commonly more expensive

units need to be available. Moreover, the limited predictability of wind generation or the

uncertainty degree in the forecasted errors calls for the need of a dynamic balancing mechanism

that is able to accomplish the required adjustments in real-time operation. Evidently, a

significantly larger share of renewables in the power system magnifies these challenges even

more. If the current electricity trading arrangements are maintained, TSOs would have to face

many more difficulties, incurring in even more inefficiencies.




From a network topology perspective, the integration of renewable sources is not evenly

distributed around the continent. While wind farms are mostly concentrated, and are expected to

further expand, in regions close to the North Sea, photovoltaic solar panels are located in

specific and sunnier areas of the continent. This uneven distribution of generation sources, if not

accompanied by a proper transmission network expansion plan, can lead to significant

modifications in the power flows through the grid, affecting in turn the current cross-border

congestion profiles between countries and underutilizing the available network capacity. An

effective congestion management approach that captures the future elements of the system and

addresses the foreseen challenges would contribute to maximizing the benefits of power trade in

Europe.

A long-term solution to the above challenges is closely linked to reviewing the current power

market design and how it manages congestion in the electricity grid. Attention should be

focused on whether it makes more economic and operational sense to enhance, from a bottom-

up approach, the current European model with additional features focused on capacity adequacy

and the expected flexibility needs or implement a new market design with top-down support.

With this approach, this thesis performs a comparative study of the two different power markets

discussed above: the zonal market with the FBMC capacity allocation method and the nodal

market.

1.5. Research objectives

Following the research problem above, the question hence arises by how much can the power

market design structure and its market features affect the provision of cost-efficient electricity in

high renewable scenarios, and, consequently, how the market participants on a European scale

are influenced by it. The research objective of this thesis can be thus formulated as gaining a

better understanding of the European power market in presence of large amounts of renewable

energy sources and under different power market designs.

The research objective prompts the following main research question:

To what extent can the power market design and the future expected large-scale integration of

renewable energy sources have an effect on the system's variable generation costs of

electricity?

Additionally, several subquestions are formulated that provide an answer to how the main

question can best be explained; to identify what is the cause of such effect, if any; and to why it

is of importance to address this research's main question, respectively. The subquestions are

formulated as follows:

1. How can the European power system be modeled to best reproduce its behaviour when

functioning under different power market designs in combination with high and low renewable

scenarios?




2. To what extent the difference in variable electricity generation costs between scenarios and

across power market designs can be attributed to an increased usage of renewable

technologies, network congestions or non-served energy costs?

3. How can the difference in electricity prices within and between European countries impact

the markets' participants behaviour in the power market?

1.6. Thesis outline and structure

This thesis covers the research objective and research questions with the following structure:

Chapter 2 presents the necessary theory of unit commitment models used in the planning and

operation of electric power systems as well as the technical and mathematical theory needed to

understand the use of optimal DC power flow models. Furthermore, the modeling objectives,

the network topology and the modeling methodology to implement the proposed power market

designs are addressed. Finally, the mathematical formulation of the specific model used is

provided.

Chapter 3 describes the renewable scenarios used based on the Adequacy Forecast & Scenario

Outlook of 2014 from the ENTSO-E, which extend to the year 2030, and the necessary input

model data and parameter assumptions made. Moreover, a section in this chapter is dedicated to

validating the output of the model with reality.

In chapter 4 the research outputs are presented and the answers to the research questions are

exposed. Here, the effect on the variable electricity generation costs of the power market design

in combination with a high penetration of green technologies is analyzed. Also, further analysis

is made on the possible cause or causes that generate the different outputs as well as the degree

and the possible implications that these outputs would have on the European energy multi-actor

context.

Chapter 5 exposes the conclusions, recommendations and some reflections. Here, the answer to

the research questions are formulated and the aim is to connect the results obtained with the

model to the current and foreseen situation in the European power system.

Some scope limitations of the thesis are explained in the following points:

A mathematical optimization model is used to model the short-term planning and

operation of the power system. Investments in new generation assets or network

capacities are not accounted for.

Rational, cost-minimizing agents are assumed and strategic bidding of market agents is

not taken into account. The optimization problem is solved from a central planner

perspective.




The optimization problem is modeled with deterministic renewable scenarios that are

based on the scenario data provided by the ENTSO-E.

Renewable generation of wind and solar assigned to every specific node of the network

is calculated based on the forecasted wind speed and solar radiation time-series in the

assumed location of that node rather than on the average of a region covered by that

node.

The flexibility of the power system has been limited by the arbitrary number of thermal

units assigned to every node of the network. This has a direct effect on the actual

commitment, startup and shutdown schedules. As an example, it is not the same having

one thermal unit of a maximum capacity of 500 MW than having two thermal units of

maximum capacity of 250 MW each. The flexibility degree with which the power

system can operate has been modified.

Hydro energy resources are left out of the scope in the generation mix in all scenarios.

In order to account for this shortcoming, demand values obtained from the ENTSO-E

database are adjusted in those countries where hydro accounts for a significant share in

the generation mix. The outputs obtained from the optimization model should therefore

be interpreted knowing that optimization of hydro resources in combination with

intermittent renewable sources will play an important role in the replacement of

expensive thermal generation thanks to their flexibility and complementary use.






CHAPTER 2

THEORY AND MODEL DESCRIPTION



Exploration of the future European Electricity Market Design Chapter 2 – Theory and Model Description


2. Theory and Model Description

Chapter 2 presents the fundamental concepts in the theory of power systems and the modeling

optimization techniques and formulations used in the field. Once these concepts and

formulations are addressed, the second part of the chapter describes the modeling approach, the

methodology and the formulation chosen for the defined problem.

2.1. Theory of Unit Commitment

Unit commitment models are needed by system operators to schedule efficiently the resources

available in order to achieve a reliable and economically sustainable operation of the power

system. Unit commitment in power systems is an optimization problem that has the main

objective of finding the minimum-cost scheduling of all the generating units in the system over

a specified period of time, usually over a short-term horizon. Simultaneously, this objective is

subject to meeting the system electricity demand, complying with different types of system

constraints (technical, operational, environmental, regulatory), and take into account the long-

term signals such as the water value in hydro energy sources and guarantee an appropriate level

of reliability.

The unit commitment problem has technical and operational constraints for every generating

unit as well as for the coupling operation of these units, and the shorter the temporal scope of

the model and the time interval used, the more detailed the modeling of the generating units

should be. In general terms, the main constraints that can be found in a common formulation of

the unit commitment problem are related to:

The minimum uptime and downtime of the units

The upward and downward ramp limits

The capacity limits of the units

The status of the units and

The reserves constraints.

The unit commitment problem, of combinatorial nature, represents a challenging but necessary

task for system operators in the daily planning of power systems. An unplanned management of

the available generation resources can cause system operators and generation companies to

incur in extremely high economic losses. A unit commitment model provides a plan of the

physical operation of the system: start-up and shut-down decisions are made and an hourly

schedule of the generation park is obtained. Moreover, this short-term planning also helps to

forecast the following:

The operational costs of the system

The generation costs

The fuel consumption

The management of the reservoirs



The utilization factors for each generation unit

The aggregated generation for each type of technology and

The marginal costs of the units.

There exist many different approaches and ongoing research in the literature on how to solve the

unit commitment problem. Several numerical optimization techniques have been used to

address the problem. These techniques include priority list methods (Sheble, 1990) (Dillon, et

al., 1978), dynamic programming (Ouyang & Shahidehpour, 1991) (Muckstadt & Wilson,

1968), integer programming (Garver, 1963) (Snyder, et al., 1987), mixed-integer linear

programming (Cohen & Yoshimura, 1983), branch-and-bound methods (Sandrin & Merlin,

1983), and Lagrangian relaxation methods (Shahidehpour & Ouyang, 1992). Priority lists is a

quick method that specifies the order in which units are switched on or off, however, the quality

of the output is not always assured. Regarding the dynamic programming technique, the precise

solution of the model can be obtained but the large computation time is the main disadvantage.

Integer and mixed-integer methods apply linear programming by relaxing its integrality

requirements but have been used only in small unit commitment problems. The branch-and-

bound method is different from other techniques since it assumes no priority ordering, it applies

a linear function to represent fuel consumption and time-dependent start-up costs and it

calculates the required upper and lower bounds. Finally, the Lagrangian relaxation technique

can give a fast solution but it might not always numerically converge providing a low quality

solution.

Among the techniques briefly mentioned above, the Lagrangian relaxation and mixed-integer

linear programming (MILP) are the most popular methods. According to (Morales-España, et

al., 2013), the world's largest competitive wholesale market, PJM, recently switched from using

Lagrangian relaxation to MILP. However, despite the many improvements made in MILP

solvers, the time needed to solve unit commitment problems continues to be a main limitation.

Therefore, a thoroughly improved MILP formulation can really make a difference in lowering

the computational overload and making it possible to run larger or more complex problems.

Reference to the formulation used in (Morales-España, et al., 2013) shall be made in following

sections.

On the other hand, a pure unit commitment model does not include the transmission network,

instead the model's approach is to consider that all generation and load is connected in one

single node. However, because generation units are scattered in different places interconnected

between them across the network, the power flows are restricted to the capacity limits of the

transmission lines. Therefore, in order to obtain a feasible commitment schedule of the

generators the transmission network should be included in the model: a transmission economic

dispatch model is needed.

A transmission constrained economic dispatch problem, namely an optimal AC power flow

model, seeks to minimize operational costs of providing electricity by taking into account the

network constraints. The model provides as results the power output of each generating unit and

the power flows through the transmission lines, as well as a specified operating reserves margin,



also known as the spinning reserve. Additionally, the model introduces the important concept of

nodal prices or nodal marginal costs which will be addressed in subsequent sections.

Finally, the merger between a pure unit commitment model and a transmission constraint

economic dispatch problem provides a full and detailed overview of the required short-term

planning carried out by a Transmission System Operator (TSO). However, there are some

drawbacks to solving the AC power flow problem. An electricity network with N nodes results

in an AC power flow with 2N non-linear equations that are solved iteratively for every time step

(Van den Bergh, et al., 2014). Given the non-linearity of the transmission constraints and the

heavy computational effort required, a simplified version of an AC power flow based on linear

programming is used in practice. Section 2.2 presents the simplifications made for the AC

power flow model.

2.2. Theory of Optimal DC Power Flow

In the following section a detailed description of DC power flows is given. The AC load flow

constraints can be formulated as linear constraints based on three main assumptions.

1. The inductive component of the AC transmission line is greater than the resistive one for all

lines ( ) which means that network losses can be ignored and the line admittances can

be simplified to line susceptances in the admittance matrix ( ). The higher the voltage

of the network in question, the more valid this assumption will be.

2. All the bus voltages have a similar magnitude in per unit values. This is achieved by

managing the reactive power in the system to maintain voltage fluctuations to the smallest

possible degree. However, real life examples show that this assumption is the largest source of

DC power flow errors.

(Eq. 2.1)

3. The phase angle differences between neighbouring node voltages ( and ) under stable

operation conditions are small. This translates into a linearization of the sine and cosine terms in

the AC power flow equations as equation 2.2 shows. In general terms, this linearization is more

correct in weakly loaded grids, but even in peak load instances, the error made by this

assumption is less than 1%.

(Eq. 2.2)

How close reality adjusts to these three assumptions is what makes the solution given by the DC

power flow model more accurate. Generally, compared to AC power flows, the accuracy of DC



power flow models when applied to high voltage grids is close to 5% when averaged over all

lines. Therefore, by keeping in mind the conclusions drawn for individual lines, the error

deviations are assumed to be acceptable.

Although a reference bus is needed as reference point in the linearization of the power flow

model, the literature (Sheble, 1990) (Baldick, 2003) (Dillon, et al., 1978) (Baldick, et al., 2005)

proves that the DC power flow formulation is not significantly sensitive to different operating

points. Therefore, as long as the network topology is kept unchanged, the DC power flow

equations can be used for all operating points indistinctly.

Figure 2.1. Impedance of line ij

In the DC power flow problem perfect voltage support, reactive power management, negligible

network losses are assumed and only line active power flows are considered. The active power

flow through an AC transmission line follows the following equation:

(Eq. 2.3)

By applying the mentioned assumptions above, equation 2.3 simplifies to the DC power flow

equation, Eq. 2.4:

(Eq. 2.4)

Similarly, the simplified DC power flow equation for nodal active power balances in one node

is shown below (Eq. 2.5), with being the positive direction of the active power flow from

node to and the susceptance of line between node and (Van den Bergh, et al., 2014):

(Eq. 2.5)



If node is now considered to be the reference bus, Eq. 2.5 can be then written in matricial

form:

(Eq. 2.6)

(Eq. 2.7)

Therefore, the voltage phase angle is calculated as in Eq. 2.8 and, by substituting this

expression in Eq. 2.4, the Power Transfer Distribution Factors (PTDFs) are obtained, as shown

in Eq. 2.10.

(Eq. 2.8)

(Eq. 2.9)

(Eq. 2.10)

Finally, the DC power flow equations of the AC transmission lines in the network are

represented in matrix form in Eq. 2.11.

(Eq. 2.11)

The Power Transfer Distribution Factors (PTDFs) linearly relate the power injections in the

nodes of the network ( ) and the active power flows in the lines ( ). In essence, the link



between and is a relation between reactances. Also called sensitivity factors, an

element in the PTDF matrix provides an approximation of how the power flow through

transmission line changes with an injection or withdrawal of one MW of active power at node

(Van den Bergh, et al., 2014). For the PTDF calculation, the reference node or slack bus needs

to be removed from the DC power flow equations in order to obtain a set of linearly

independent equations.

Finally, the active power flows are bounded by the capacities of the transmission lines, Eq. 2.12.

(Eq. 2.12)

Until this point, only AC transmission lines have been considered in the formulation. However,

High Voltage Direct Current (HVDC) lines can also be part of the meshed high voltage grid. In

case the HVDC line is embedded within an AC grid, the power flow through this line can be

fully controlled. If the HVDC line connects extremes of the network, like for example offshore

wind farms with the continental grid or interconnects two asynchronous power systems, then the

HVDC transmission line is not a true power flow controlling device.

Van den Bergh (2014) proposes a methodology to model these lines based on the replacement

of the line capacity for a combination of a positive and negative power injections at the two

nodes that connect the HVDC line. The methodology applied in this research uses Van den

Bergh's idea of replacing the power flow through the HVDC line ( ) for the nodal injections

in the corresponding nodes ( ) through the relationship given by , the incidence matrix

of the HVDC network. is a L N matrix with if line L starts at node N,

if line L ends at node N and if line L is not incident to node N. Therefore,

represents a vector that contains the nodal injections in every node due to the power flow

through the HVDC lines (Eq. 2.13).

(Eq. 2.13)

Even though no reactance is associated to the HVDC lines, the DC power flow equations of the

AC lines should also include the effect of the power flows in the HVDC lines. This effect is

included in Eq. 2.14 by adding the term to Eq. 2.11. Finally, Eq. 2.14 are the DC power

flow equations of the AC transmission lines in a network with HVDC lines.



(Eq. 2.14)

Equation 2.14 implicitly takes care of the power balance in the nodes that are connected to at

least one AC transmission line. However, an additional constraint needs to be added in order to

guarantee power balance in the nodes connected only to HVDC lines (Eq. 2.15). This constraint

includes the net injection calculated as a result of the net generation, demand and energy not

served in every node ( ) plus the net inflow due to the power flow of HVDC lines

connected to node

(Eq. 2.15)

2.3. Locational Marginal Prices

The Locational Marginal Prices (LMP), or nodal prices, are also the marginal costs of

electricity. Nodal prices guarantee at every point in time that demand is met at minimum cost

and price differences reflect generation and transmission capacity scarcity, as already explained

in section 1.3. The LMP mechanism takes into account all the physical transmission constraints

in the clearing of the day-ahead market. Based on a real representation of the network and a

transparent approach, the price calculation is done a posteriori, after solving the linear

programming model, and is based on the dual variables of the system balance and network

constraints.

Every constraint in the model has a corresponding dual variable, also called shadow price. A

dual variable provides crucial information about the system. It measures how much the

objective function varies when the corresponding constraint is incremented by one unit. In other

words, in this case they measure the change in total variable generation costs when demand,

generation or non-served energy vary by one unit.

It should be noted that non-binding constraints correspond to shadow prices equal to zero. In the

event that some constraints are binding, for example a specific line is operating at its limit

capacity, the shadow price for that line will be non-zero and thus it indicates the presence of

transmission congestion in the network. The system balance constraint, specified as an equality

constraint (Eq. 2.16), always binds; it suggests a price that always supports balance in the

system.

Eq. 2.16. System balance constraint and its dual variable.



The following fundamental formula, in the absence of network losses, is used to calculate the

LMP in any node:

Eq. 2.17. Formula of locational marginal prices (€/MWh) based on dual variables.

where:

Node

Hourly period

Shadow price (dual variable) of the system balance constraint

AC power transmission line

– Power Transfer Distribution Factor of AC transmission line due to node

Shadow price (dual variable) of the network constraint of AC transmission line

Shadow price (dual variable) of the power balance constraint in nodes with only

HVDC lines

The advantage of using a nodal pricing mechanism over a FBMC or ATC mechanism is that it

efficiently manages congestion in the network without interference of the TSOs. This is because

they internalize the energy losses and congestions in accordance with the network constraints.

As an important result, nodal prices give efficient economic short-term signals to market

participants and stimulate the required investments and the appropriate performances of system

operation.

2.4. Modeling approach

Large-scale integration of renewable energy sources in the near future is expected to become a

main driver to enhance operational practices to provide a more flexible and efficient functioning

of the European power system. This flexibility can be achieved from the generation side, from

both conventional and renewable generation, by improving their system reliability and stability

capabilities; it can also be achieved from the load side, with demand response and also from the

operational side, by better exploiting the existing transmission infrastructure and by sharing

greater cooperation between regions for transmission expansion.

Moreover, the power market design, which in general terms dictates the rules of the power

system's functioning, has an impact on how the power system's flexibility is managed and

therefore it also has an effect on the variable generation costs of the power system. In order to

stress the relevance of the power market design in presence of increasing renewable capacity, an

energy model is developed. The following subsections elaborate on the modeling objectives, the



type of the chosen model and its scope and the modeling methodology used to implement the

proposed power market designs.

2.4.1 Modeling Objectives

The interest of this research lies in gaining further insights on the functioning of the European

power system under different power market designs taking into account the large amounts of

renewable sources that will be integrated in the following years. Up until now system operators

have had relatively little problems integrating small amounts of RES. However, according to the

European Commission, a new renewable energy target has been set: at least 27% of the final

energy consumption in the European Union as a whole should come from renewable sources by

the year 2030. This is expected to unfold many challenges for TSOs in efficiently managing the

network as explained in section 1.2.2.

In order to reflect the impacts of the high RES integration in the functioning of the power

system, a bottom-up model approach is chosen. With this approach the considered energy

system is technically and economically parameterized and looked at from a technological

perspective. Following this, a comprehensive analysis of the technological aspect under study is

allowed (Jägemann, et al., 2013), in this case the large deployment of RES technologies in the

power system under different power market designs.

Within this bottom-up approach, an optimization model is selected over a simulation model, like

agent-based models or system dynamics models. TSOs are the entities that operate and manage

the power system, and their aim is to manage it in the least costly way possible. Hence, the

ability of the optimization model to determine a system optimal solution from a centralized

perspective makes it more suitable for the present research.

The aim is to analyze the impact of the power market design on the power system's variables,

like total variable generation costs, hourly electricity prices, hourly line congestions, RES

curtailment and total production derived from different fuel types, while subjected to several

scenarios of increasing degree of RES integration. Therefore, it is preferred to develop a model

that gives the optimal functioning of the system for a given objective function, i.e. minimization

of total variable generation costs, and a set of constraints that express the technical limitations

and political targets in the system.

2.4.2 Modeling the Day-Ahead Market

With the theory of unit commitment and optimal DC power flow models explained in sections

2.1 and 2.2, the optimization model developed in this research solves the unit commitment

problem and a transmission constrained economic dispatch of the units for a conceptual network

of North-West and central Europe. Figure 2.2 shows a representation of the modeled network.



The model simulates the day-ahead market for the European network in Figure 2.2 from a

centralized decision making point of view. This means no strategic behaviour between agents is

considered in the model. The day-ahead market is solved by minimizing total variable

generation costs at the same time it allocates the hourly demand over the existing generating

units and complies with the technical limitations of the power system, i.e. maximum capacities

of the generators or transmission line limits.

Morales-España et al. (2013) proposes a mixed-integer linear programming (MILP) formulation

of the unit commitment problem. Compared to other existing and efficient formulations, this

tighter and more compact formulation reduces the feasible region that needs to be explored by

the solver and it increases the speed with which solvers search the optimal integer solution in

that feasible region. This model's formulation characteristics lower the model's computational

complexity in comparison to other unit commitment formulations, making it attractive for large

unit commitment problems representing large power systems. For this reason, this UC

formulation is chosen in this research. For a detailed elaboration on the peculiarities and

advantages of the mentioned formulation, the interested reader is referred to Morales-España et

al. (2013). Additionally, the network elements have been added to the UC formulation as it will

be explained in the mathematical formulation section, section 2.4.

Figure 2.2. Conceptual European network modeled in the day-ahead market



Technological scope

The model includes the operational costs and technical characteristics of 873 conventional

power plants as well as their commitment decisions. The following generation technologies are

considered:

Nuclear power plants

Coal-fired steam power plants powered with coal or anthracite

Lignite-fired steam power plants

Heavy and light fuel oil power plants

Combined cycle gas turbines (CCGT) powered with natural gas

Gas-fired steam turbines

Regarding renewable technologies, both onshore and off-shore wind and solar energy

production are included in the generation mix. Normalized generation profiles of wind and solar

energy which account for the characteristic wind and solar daily patterns are included as input

data in the model from The National Centers for Environmental Prediction (NCEP) Climate

Forecast System (CFS) database (Saha, et al., 2011), and scaled with the renewable installed

capacity in every RES scenario considered. The thermal generation mix is also updated with the

scenarios, except hydro generation, which given the lack of input data, was not considered.

It should be noted that renewable energy production is given priority in the model as no

operational cost is assigned to it and curtailment, this is disconnecting renewable sources from

the power grid, is allowed to happen in situations of sudden lack of demand or line congestions.

Temporal scope

The model calculates the optimal economic dispatch schedule that meets the hourly demand for

one week in July in a reference scenario in the year 2014. Furthermore, another five scenarios

representing different projections of RES integration for the years 2020 and 2030 are used to

help assess the system's operation adequacy in these situations. These scenarios are named as

scenario EU 2020 and Visions 1, 2, 3 and 4 for the year 2030. A detailed description of these

scenarios is done in chapter 3.

Regional scope

Twelve countries in the European Union are covered in the modeled network with one, two and

up to five nodes per country, in the case of Germany. In total the model contains 22 nodes

connected by a total of 72 transmission corridors. Figure 2.2 shows the network topology with

the nodes and transmission lines used. This power system is a highly simplified version of the



real high voltage European network. However, the number and type of connections between

countries are defined in accordance to reality.

Transmission corridors in the network differ between AC and HVDC lines. Their transmission

capacities and number of circuits in each corridor are defined based on (Van Blijswijk, et al.,

2015). The HVDC links mainly connect remote regions with renewable resources to load

centers in the meshed AC continental power system. Moreover, both types of transmission

corridors are modeled differently. As already mentioned in section 2.2, AC transmission lines

have an electrical reactance associated to them which largely determines the distribution of the

power flows in the grid. On the other hand, the flow through HVDC transmission lines is not

dependent on an electrical reactance but still affects the flows in the rest of AC transmission

lines in the network. The specific network formulation will be detailed in section 2.5.

Regarding the allocation of thermal generation in the grid, a generation database from TU Delft

(Enipedia) was initially used as an indication of the number of existing units and their location

in every country. However, given the incompleteness of the data, an arbitrary estimation of the

number of generating units in every node and for every type of technology was made. This

estimation was calculated by taking into account the country's generation mix for the scenario in

question and assuming a range of realistic capacity values of the generators according to the fuel

type. No change in the number of units is assumed across scenarios, only the capacities of these

units change in order to adjust to the new generation mix in the different scenarios. Tables D.6

to D.11 in Appendix D show the capacity values used according to the fuel type and the

scenario.

With respect to renewable generation, the normalized wind and solar feed-in profiles implicitly

account for regional variations of wind speed and solar radiation throughout the power system.

Regarding demand, the national hourly demand value obtained from the online Transparency

Platform of the ENTSO-E is allocated to every node in the country proportionally to the

population in every zone covered by every node. National population statistics from the year

2013 are mostly used. In this way, every hour and every node in the network have one demand

value assigned.

2.4.3 Model Implementation of Power Market Designs

In this section the methodology used to model the different power market designs discussed in

section 1.3, the FBMC allocation method for the zonal market and the nodal market, is

presented. They are modeled according to how congestion is managed in each case. Figure 2.3

shows a schematic representation of the modeling process described below.

In the nodal market, as previously explained congestion is entirely handled through the market.

In this case the unit commitment and economic dispatch model is run with all the network



constraints activated. The output dispatch of the model is therefore feasible with all the

constraints and needs no adjustments.

Figure 2.3. Schematic representation of the modeling process of the power market designs

A zonal market with a Flow Based Market Coupling (FBMC) capacity allocation mechanism is

modeled in two different steps. In a first phase the unit commitment schedule of the generating

units and the economic dispatch is calculated by the TSO once the clearance of the market is

done. In the model this is done by running the optimization algorithm taking into account only

cross-border interconnections. The national transmission lines are ignored by setting their

inductive reactances to zero and their capacities to a very large value.

During the second phase, the TSO evaluates intra-zonal congestion within national boundaries,

and checks the feasibility of the dispatch outcome with the national transmission constraints. In

practice, this translates into running the MIP economic dispatch model once again taking into

account all the transmission lines. In the redispatch phase the commitment, startup and

COPPER PLATE MARKET

Run the Model:

All interconnections ignored

System generation costs obtained

NODAL MARKET

Run the Model:

All interconnections considered


Nodal prices obtained

ZONAL MARKET

REDISPATCH PHASE

Run the Model:

Commitment decisions maintained

All interconnections considered

System redispatch if necessary


Nodal prices obtained

ZONAL MARKET

FIRST PHASE

Run the Model:

Interconnections within

zones ignored

Commitment decisions of

units obtained



shutdown schedules obtained from the economic dispatch in the first phase are fixed; no startup

or shutdown of plants can now take place. The redispatching process should be facilitated by

those units who are already committed.

Moreover, given the direction towards where Europe is heading, a reference model is used for

comparison purposes. This model, referred to hereafter as Copper Plate, assumes a perfectly

future integrated Internal Electricity Market with the sufficient network capacity in all locations

that ensures a single wholesale price and prevents any congestion from existing in the whole of

Europe. In practice, this major assumption translates into running the unit commitment and

economic dispatch model without taking into account the transmission constraints of the

European network. Knowing that the current situation is far from such ideal model, it serves as

a reference with respect to the zonal and nodal markets and it provides with a rough estimation

of the studied parameters in a yet unreal electricity system.

2.5. The Optimization Problem. Model Formulation

The clearance of the day-ahead market is modeled with a unit commitment and a transmission

constrained economic dispatch model (UC & TCED). The optimization model minimizes the

total variable generation costs of the considered power system (Morales-España, et al., 2013).

These costs include fixed generation costs comprised by the fixed fuel consumption costs and

the startup and shutdown costs, and the operational costs that include the variable production of

the units and the CO2 emission costs.

Simultaneously, the optimization problem is constrained to a set of technical limitations of the

thermal units, to maximum wind and solar hourly generation profiles and to line transmission

capacities, among the most relevant ones.

The optimization problem is programmed using GAMS (General Algebraic Modeling System),

a high-level modeling software for mathematical programming and optimization. By using the

GAMS solver, Cplex, large linear and mixed integer programming problems can be solved. The

GAMS code of the model can be found in Appendix A. An interface with Excel-GAMS-Matlab

was prepared to input the data into the software and generate the output figures.

The mathematical formulation implemented in the model will be presented and explained in the

section below. Given its complexity, a text representation of the model is presented in Text Box

1 to help the reader understand the functioning of the model. The tight and compact formulation

used is adapted from (Morales-España, et al., 2013), and additional parameters and variables

have been added in order to model the dispatchable HVDC transmission lines existing in the

network.

The code in GAMS is divided mainly into four differentiated sections: sets, parameters,

variables and equations. In the following section the equations will be presented and explained.

Refer to the Nomenclature for a clarification of the defined sets and parameters used in the



model. For consistency reasons, uppercase letters denote parameters and sets. Lowercase letters

denote variables and indexes.

Text Box 1. Text scheme of optimization model

Optimization Problem:

Unit Commitment & Transmission Constrained

Economic Dispatch

Minimize

Total variable power system variable generation costs to supply electricity

demand for a week

Fixed fuel cost consumption

Start up and shutdown cost

Variable production and operation & maintenance cost

CO2 emission cost

Subject to

Power system balance

Total thermal energy production + total renewable energy

production + total non-served energy in the system needs to be

equal to the total energy demand in the system in every hour

The provision of upward and downward operating reserves for

every hour in the system needs to be ensured to respond to

unexpected events in the grid. These reserves are estimated to be a

percentage of the hourly demand and are provided by the online

generating units that can increase or decrease their output over their

scheduled output in limited time. For simplicity reasons, reserves

have been assumed to be zero.

Constraints of thermal generating units

Total production of committed units. If a generating unit is

committed, its production needs to be above the unit's technical

minimum output.

Generator output must remain below its maximum capacity.

Logical commitment constraint. A unit can start up, be committed

and shut down in different, consecutive or alternate, periods but it

can never start up and shut down simultaneously.

Minimum time online and offline. If a unit is committed, it must

remain online for a minimum number of hourly periods. Similarly,

if a unit is shut down, it must remain offline for a minimum number

of hourly periods.

The production above technical minimum of the unit + upward

reserve of the unit needs to be less or equal to the difference

between the unit's maximum capacity and its technical minimum

output.

The downward reserve of the unit needs to be less or equal to the

production above technical minimum of the unit.



Ramping limits of the units. The increase or decrease in a unit's

output has to remain lower than its maximum upward and

downward ramp capabilities, respectively.

Network constraints

Capacity limits of the transmission lines. The flow of energy

through the AC and HVDC transmission lines needs to remain

lower than the maximum capacity of these.

Power flow in AC transmission lines. The power flowing through a

specific AC transmission line is equal to the net power injection in

a node multiplied by a factor (PTDF), and summed over all the

nodes in the network that connect to any AC line.

The net power injection in a node includes production + non-served

energy demand + the net power inflow from HVDC lines

connected to that node.

The factor PTDF relates the net power injection in a node that

connects to any AC line, to the flow of the AC line of whose power

flow is being calculated.

The power balance in every node connecting to any AC line is

implicitly guaranteed with this equation.

Power balance in nodes connecting only to HVDC lines.

Production + non-served energy demand + the net power inflow

from HVDC lines connected to that node should be equal to zero.

2.5.1 Objective Function

The unit commitment model's algorithm seeks to minimize the total variable generation costs to

supply electricity for one week in the simplified European power system considered. In general

terms, these total variable generation costs are defined as the sum of: (i) the fixed fuel

consumption costs , (ii) the variable costs of production (ii) the startup and

shutdown costs and , (iii) the non-served energy costs and (vi) the CO2 emission

costs , as shown in equation 2.18.

(Eq. 2.18)

The fixed fuel consumption cost is the sum over all the generators and all hourly periods of

the fixed cost of fuel in €/hour incurred for all thermal generating units (Eq. 2.19). This cost is

obtained based on the fixed fuel consumption of the units according to the fuel type used. Units

that have lower fixed fuel consumption rates like nuclear fuel, lignite, coal and anthracite units,

have a lower fixed fuel cost compared to those units using natural gas, gas or light and heavy

fuel oil.



(Eq. 2.19)

The short run marginal cost (SRMC) of production refers to the marginal cost of the generators.

It is calculated over all generators and all hourly periods (Eq. 2.20). It includes the

production cost and the operation and maintenance cost of the units. The production cost can be

approximated to the fuel cost in €/MWh divided by the operating efficiency of the generating

unit.

(Eq. 2.20)

Startup costs reflect the consumption of fuel required to reach the optimal conditions of

temperature and pressure in the boiler of the generating units. This cost is calculated over all

generators and all hourly periods following equation 2.21, where represents the startup

cost of unit per capacity installed, is the unit's maximum capacity and is the unit's

binary startup variable. In this model, shutdown costs have been assumed to be a 10% of the

startup costs.

(Eq. 2.21)

(Eq. 2.22)

In general terms, the cost of non-served energy is a result of the total amount of energy not

supplied in every node and hour, , multiplied by the cost of non-served energy in

€/MWh calculated over all the nodes and all hourly periods as shown in equation 2.23.

(Eq. 2.23)

The objective function also includes a term for total CO2 emission costs for those thermal units

with an emission rate (tonnes of CO2/GWh). The emission rate multiplied by the cost of CO2

emissions, (M€/ tonnes of CO2), by the power output of the unit and summed over all

thermal generators and all hourly periods gives the total CO2 emission cost (Eq. 2.24).



(Eq. 2.24)

2.5.2 Constraints

Alongside the objective function several technical constraints exist and they need to be

complied with. The following section is devoted to review these constraints.

2.5.2.1 Power system balance requirement.

An energy balance is required between demand and thermal ( and renewable

production at all time periods in the power system. Curtailment of renewable generation is

allowed in the model if it is required to comply with system balance. Additionally, a variable for

non-served energy is included to take into account situations in which there is a lack of energy

production or the network limits the supply of electricity to a particular load center. This is

guaranteed with equation 2.25.

(Eq. 2.25)

2.5.2.2 Production of committed units.

The total production of a unit is modeled in two separate terms, as shown in equation 2.26. On

the one hand there is the technical minimum output of the unit once it is committed and on the

other hand a term is included in case the output is above that minimum.

(Eq. 2.26)

Moreover, the initial behaviour of the units is limited by their initial conditions (Morales-

España, et al., 2013). In practice, a day in advance is modeled in order to obtain the initial

conditions for the following day. If the initial output of the unit lies above its technical

minimum output then the unit is committed.



2.5.2.3 Logical Commitment and Minimum Up and Downtime constraints

The following constraint (Eq. 2.27) guarantees that the commitment, startup and shutdown

binary variables take the adequate values between hourly periods.

(Eq. 2.27)

On the other hand, equations 2.28 and 2.29 refer to the minimum number of hourly periods that

the units must be online and offline. As specified in (Morales-España, et al., 2013) the

combination of these constraints results in , and this assures that a unit does not

start up or shut down simultaneously.

(Eq. 2.28)

(Eq. 2.29)

2.5.2.4 Generation capacity limits

The proposed formulation from (Morales-España, et al., 2013) for the generators' capacity limits

over their production and their upward and downward spinning reserve is constrained by the

committed, start up and shutdown status of the units. It is realistic to assume that, immediately

after starting up a unit or immediately before shutting down the unit, this one is producing at its

minimum level. Equations 2.30, 2.31 and 2.32 assure that, both if a unit is started up in hour

( ) or shut down in ( ), the production of that unit in hour will be its

minimum output ( ) since the right term would be multiplied by zero and

is defined as a positive variable.

(Eq. 2.30)

(Eq. 2.31)



(Eq. 2.32)

(Eq. 2.33)

It is noted that equations 2.30 and 2.31 are applied for the subset which contains the

generating units with = 1. On the other hand, equation 2.32 is included for those units with

, this is when the units are operating for at least two online periods. Although equation

2.30 and 2.31 are feasible for both when and when , using the combination of Eq.

2.30, 2.31 and 2.32 provides a tighter and more compact formulation. The reader can refer to

(Morales-España, et al., 2013) for more information about the use of this formulation.

2.5.2.5 Upward and downward system operating reserves

The following constraints (Eq. 2.34 and Eq. 2.35) ensure the provision of upward and

downward operating reserves for every hour in the system. These reserves, provided by the

online generating units, are defined as the power that these units can increase or decrease over

their programmed output within a limited time in response to an automatic generation control

(AGC). For simplicity reasons, reserves are assumed to be zero.

(Eq. 2.34)

(Eq. 2.35)

2.5.2.6 Ramping limits

Equations 2.36 and 2.37 ensure that the upward and downward ramp rate limits of all the units

are respected.

(Eq. 2.36)

(Eq. 2.37)



2.5.2.7 Network constraints

The power flow in AC transmission lines is calculated using a linearized approximation of the

AC power flow problem: the DC power flow model. The Power Transfer Distribution Factors

(PTDF) are calculated as the result of the linearization process. This simplification is based on a

series of assumptions previously explained in section 2.2.

For every hourly period and for every AC transmission line in the network, the active power

flows in these lines are calculated using the linear relationships (PTDFs) between the active

power flows and the net power injections in the subset of nodes that connect at least one AC

transmission line ( (Eq. 2.30). These net power injections are obtained from the power

balance in every bus; it includes the total thermal and renewable generation, the demand, the

non-served energy and the injections due to the power flows in the HVDC lines connected to

the node in question. Equation 2.30 implicitly ensures the power balance in those nodes that are

connected to at least one AC transmission line.

(Eq. 2.30)

Regarding the power flows in the HVDC lines ( ), these are calculated through the power

balance equation of those nodes in the network connected only to HVDC lines (Eq. 2.31).

(Eq. 2.31)

Moreover, Eq. 2.32 and Eq. 2.33 ensure that the transmission line capacities are respected in

both AC and HVDC lines.

(Eq.

2.32)

(Eq.

2.33)





CHAPTER 3

SIMULATION SETUP



Exploration of the future European Electricity Market Design Chapter 3 – Simulation Setup


3. Simulation Setup

Until now, the focus has been set on the theory and application of unit commitment and

economic dispatch models for the present research and to the approach used to model the

proposed power market designs.

The following chapter discusses the renewable scenarios used and the assumptions made in the

optimization model that will serve to analyse how well the different power market designs

accommodate the different degrees of renewable energy sources integrated in the power system.

Moreover, sections 3.2 and onwards give a description of the input data and its sources.

Demand, operating reserves, the properties of the generators and their costs, renewable

generation and the parameters of the network are identified and used in the formulation

previously presented.

3.1. Scenario definition

Five future scenarios with different degrees of renewable energy integration in Europe are used

to help assess the degree of uncertainty that a long-term generation adequacy assessment entails.

Moreover, these scenarios help evaluate the evolution of key parameters relevant to a proper

functioning of the European interconnected power system. Trends on nodal electricity prices,

the degree of power line congestions, curtailment of renewable generation and the evolution of

the system's total variable operation costs over the three proposed power market designs is the

main focus of the scenario analysis.

The scenarios are based on the Scenario Outlook & Adequacy Forecast (SO&AF) from 2014 to

2030 (ENTSO-E, 2014a), which are annually published by the ENTSO-E, as part of the Ten-

Year Network Development Plan (TYNDP) (ENTSO-E, 2014b).

The Scenario Outlook & Adequacy Forecast report sets out three scenarios (Scenario A

"Conservative", Scenario B "Best Estimate" and Scenario EU 2020) and four visions for 2030.

In these scenarios generation and demand forecasts are specified for all Member States and the

adequacy of the ENTSO-E's interconnected power system is assessed on the mid and long-term

horizon focusing on power balance, margins, energy indicators and the generation mix. Thus, it

provides stakeholders in the electricity sector with what the future holds on the European level.

The specific scenarios used in this research are the following:

Scenario B "Best Estimate", as the reference scenario for year 2014.

The EU 2020 Scenario.

The four Visions for 2030.

These scenarios are non-probabilistic, instead they have associated estimations of the generation

mix and the demand profiles. The details on the assumptions on which these scenarios are built



are shown in the next subsections. Figure 3.1 gives an overview of the generation mix for the

considered network across the chosen scenarios.

Figure 3.1. General overview of the generation mix across all scenarios.

Source: Own source, calculated with the input data in the model.

In relative terms, it can be noted that the share of installed renewable capacity in the entire

power system considered is expected to increase from 26% in 2014 to 61% in year 2030 while

all the thermal technologies decrease in share. Along with this general decrease of thermal

power plants, the lowest capacity values of coal-fired steam power plants are spotted in Vision 3

(51,7 MW). This is combined with the highest capacity values of gas-fired steam power plants

and combined-cycle gas turbines in Vision 3, which increase by 27% with respect to the

capacity installed in scenario EU 2020 (see Table B.2 in Appendix B). Therefore, there is a

clear shift expected towards renewable sources and, consequently, a need for flexible responsive

technologies to react in case of sudden lack of wind or sun.

3.1.1. Reference scenario

Scenario B "Best Estimate" is used as the reference base case scenario to model the current

situation in the year 2014. It is a bottom-up scenario based on the TSOs' expectations of

potential developments for the period 2014 to 2025 given that market signals incentivize

investments.



In terms of generation mix, this scenario takes into account the generation capacity considered

to be as sure and whose commissioning is under process as well as the future commissioning of

power plants considered as credible according to the TSOs' available information. The request

to obtain connection to the grid by producers is not the only source of estimation used by the

TSOs to consider the likelihood that a project takes place, but also the regional economic

assessments of these are taken into account.

Figure 3.2 shows the generation mix per country used to model the network in year 2014 (See

associated Table B.3 in Appendix B). The share of renewable installed capacity is well below

30% except in countries like Germany, Norway or Denmark, which already have significant

amounts of solar plants and wind farms, respectively. For further political and economical

viewpoints considered in this scenario the reader can refer to the Scenario Outlook & Adequacy

Forecast 2014-2030 (ENTSO-E, 2014a).

Figure 3.2 . Breakdown of generation mix per country in Scenario B year 2014.


3.1.2. Scenario EU 2020

Scenario EU 2020, built from a top-down approach, is derived from the National Renewable

Action Plans (NREAP) of the member states or from the equivalent governmental plan in case

of lack thereof. It gives an estimation of the required future renewable energy developments to

comply with the European 20-20-20 objectives. Conventional generation capacity forecasts that

support the European renewable plan are also envisaged on a national basis. Moreover, the



assumptions in this scenario also serve as an important tool for the ENTSO-E's analysis related

to the requirements for future grid development.

Compared to Scenario B, the difference of both scenarios is reflected between the number of

projects that are likely to happen given the regional market incentives with the number of

projects needed to meet a set of political objectives on a European level. In this respect, and

provided that all countries meet their national renewable targets, Figure 3.3 shows that almost

all countries increase their renewable capacity share to at least 30%. (See associated Table B.4

in Appendix B).

Simultaneously, there is a considerable decrease in the system of coal-fired power plants (lignite

and hard coal) (12%) and fuel oil plants (27%) with respect to the reference scenario. Nuclear

power plants share drops by a 4,5% while gas-fired plants and CCGTs slightly increase their

share by 3,8%. For further political and economical viewpoints considered in this scenario the

reader can refer to the Scenario Outlook & Adequacy Forecast 2014-2030 (ENTSO-E, 2014a).

Figure 3.3 . Breakdown of generation mix per country in Scenario EU year 2020.


3.1.3. The Visions for 2030

The year 2030 serves as a bridge between the European energy targets of 2020 and 2050. The

four Visions presented in the ENTSO-E Scenario Outlook & Adequacy Forecast are

conceptually different from the scenarios shown until now. There are no probabilities associated



to them but they are also not forecasts. Taking into account the high uncertainties of such long-

term predictions, quantitative data is gathered for the four Visions based on public economic

analyses, existing European documents and previous ENTSO-E market studies, such as the Pan-

European Market Modeling data used for the Ten Year Network Development Plan (TYNDP),

2014.

The aim of these Visions is to estimate the extreme values between which the actual evolution

of parameters in the system is expected to lie (ENTSO-E, 2014a). It differs from the objective of

scenario B and scenario EU 2020, which estimate the actual evolution of parameters under

different assumptions.

The Visions are named as following:

Vision 1 - Slow Progress

Vision 2 - Money Rules

Vision 3 - Green Transition

Vision 4 - Green Revolution

Source: TYNDP, (ENTSO-E, 2014b).

Underlying every Vision there are significantly different assumptions. Visions 1 and 3 are built

from a bottom-up approach with each country's energy policy and assume thereby a weak

integration of the European energy market, with Vision 1 assuming an overall delay in the

energy goals of 2050 and Vision 3 considered to be on track with these policy goals.

On the other hand and in contrast to the above, Visions 2 and 4 are built from a top-down

approach, and assume a strong integrated European market. The generation mix development

strategy is derived from Visions 1 and 3 after a harmonization of the data from all countries.

Figure 3.4 and Figure 3.5 give a brief overview of the political and economical frameworks and

the generation and demand assumptions under which the Visions are built, respectively. For

further details on these assumptions, the interested reader is referred to the Scenario Outlook &

Adequacy Forecast 2014-2030 (ENTSO-E, 2014a).



Figure 3.4. Political and economic frameworks of the four Visions. Source: TYNDP 2014.

Figure 3.5. Generation and load frameworks of the four Visions. Source: TYNDP 2014



Figure 3.6 and Figure 3.7 show the breakdown of the generation mix per country for Vision 1

and Vision 4. Appendix B contains the analogous figures for Visions 2 and 3, and the associated

Tables Table B.5 to Table B.8 in Appendix B.

Figure 3.6 . Breakdown of generation mix per country in Vision 1 year 2030.


Figure 3.7 . Breakdown of generation mix per country in Vision 4 year 2030.




3.2. Model Data

The optimization model used to model part of the European power system is very data intensive

due to the large number of parameters required related to generating units and transmission

lines. The next subsection describes the input parameters in the model, the assumptions adopted

and the sources used.

3.2.1. Demand

In terms of load forecast, the best national estimate by the TSOs is used. The load estimation is

based on demography, economic growth and energy efficiency policy assumptions.

Hourly load values in MW for all the countries in the study were collected for the second week

(from day 7 to day 14) of January and July 2014. These demand time series were obtained from

the ENTSO-E consumption data portal. These two different weeks in the year were selected in

order to account for seasonal differences in the demand profile and in the type of technologies

used.

The year 2014 was chosen for the reference scenario presented in section 3.1 since it was the

latest available online data and it fits with the scenario B for 2014 from the ENTSO-E. The

same time series from 2014 is used for the scenario EU 2020 and it is scaled by applying to it an

annual percentage demand increase (see Table C.1 in Appendix C). As for the Visions for 2030,

the time series obtained from the TYNDP (ENTSO-E, 2014b) are used instead.

Given that in the considered network in some cases there are multiple nodes per country, the

approach followed to allocate the total hourly demand value of each country is to divide this

value of the demand proportionally to the population in the region covered by every node. In

this way, demand values are obtained for every hourly period and every node of the network.

Population data is retrieved from national statistics websites of the different countries. Table C.4

in Appendix C shows the criteria used to divide the country into their corresponding number of

zones and assign them to each and every node.

On the other hand, in the case that the hourly demand in any node cannot be fully met due to

transmission constraints or lack of generation capacity, a high penalization cost is introduced in

the objective function of the optimization problem, as already stated in section 2.5.1, in order to

minimize this term as much as possible. The cost parameter used for the non-served energy cost

amounts to 200 €/MWh. This magnitude is higher than the most expensive generator in the

system, with a marginal cost of 111 €/MWh, and it lies in the range of the marginal costs of

peak plants in balancing markets. In case there is a large unbalance in the system, these power

plants will likely be called upon in the balancing market to meet these system unbalances.



3.2.2. Generator properties

Generating units have three different types of properties: cost properties, technical

characteristics and properties related to their initial operating conditions. In the subsections

below follows a description of the assumptions made regarding the required input parameters of

these units.

3.2.2.1. Generator costs

Financial characteristics of generators depend on the type of technology in question. Since the

model is focused on the short-term planning and schedule of the units for the day-ahead market,

fixed costs and capital investments are not included. From among a unit's variable costs, it is

relevant to differentiate between the following cost parameters:

Fuel cost

Fixed fuel cost

Operation & maintenance cost

Short-run marginal cost

Startup & shutdown cost

It is not the same to operate a nuclear power plant or a gas-fired steam turbine based power

plant. The plant's flexibility in terms of response reaction to startup and shutdown commands is

directly connected to its operational costs. A nuclear power plant is less flexible and has lower

operational costs.

Fuel cost

According to the type of technology employed, the fuel used will be more or less expensive.

This is reflected in this parameter which defines the cost in Euros of fuel used per MWh of

energy produced of every thermal generator. Average values are obtained from (Neuhoff, et al.,

2013) and are shown in Table D.1 in Appendix D.

Moreover, some random variability is introduced in the fuel cost of every generator

understanding that the cost of every technology will not be the same for all generators spread in

the network. Therefore, a random maximum value of +/- 10 % of the average value is added to

the input parameter.

Fixed fuel cost

This parameter refers to the cost of the fixed amount of fuel that is required or consumed during

the time that the thermal generator is turned on. Again, it depends on the fuel type of the

generator and it is measured in Euros per hour. Figure 3.8 shows the linear approximation of the



input-output characteristic of a generic generator . The term represents the fixed fuel

consumption in thermie per hour and the variable consumption in thermie per MWh. See

Appendix D for more information.

Figure 3.8. Linear approximation of input-output characteristic of generator g.

Source: Adapted figure from Comillas University.

Operation & maintenance cost

According to the type of fuel, this parameter is defined as the cost to operate and maintain a

generating unit in Euros per MWh of energy produced. In a similar way, and in order to

facilitate the algorithm to choose the most cost-effective generators, a variation of +/- 10 % is

introduced with respect to the average cost shown in Table D.3 in Appendix D.

Short-run marginal cost (SRMC)

This parameter corresponds to the overall variable cost of a generator and it equals:

The average operating efficiency values of the different types of technologies are shown in

Table D.4 in Appendix D and are taken from (Neuhoff, et al., 2013).



Startup & shutdown costs

The cycling process of thermal power plants is becoming more and more relevant in power

systems with increasing penetration of renewable sources since it is closely linked to the

operational flexibility of the thermal plants. Starting up and shutting plants down has a

degenerative effect on the units' components due to sudden variations in temperature and

pressure accelerating failures and outages (Van den Bergh & Delarue, 2015). Therefore, the

power plant cycling process has associated startup and shutdown costs.

The startup cost in this model has been defined as the cost in million Euros per GW of installed

capacity of starting up the generator unit from a "cold" status. Additionally, startup costs consist

of direct and indirect startup costs. Table D.5 in Appendix D shows the adopted startup costs in

the model, taken from (Van den Bergh & Delarue, 2015).

Direct startup cost (€/MW) - Reflects the cost of fuel consumed needed to reach the

optimal conditions of temperature and pressure in the power plant's boiler, as well as the

CO2 emissions and auxiliary services during the startup process. In reality this cost is a

function on the time the unit has been shut down. For simplicity reasons, it is assumed

this is constant.

Indirect startup cost (€/MW) - Incorporates the long-term cycling costs such as the

capital replacement and maintenance costs into a short-term operational horizon.

Regarding the shutdown costs, this cost is incurred due mainly to the waste of fuel when the

thermal generator has been disconnected from the grid. Besides, it can represent the wearing off

cost due to startup and shutdown operations that reduce the plant's lifetime. A 10 % of the

startup cost is considered for this parameter.

3.2.2.2. Technical and environmental characteristics

Thermal generators' technical characteristics such as the upward and downward ramping limits,

the minimum technical output, the minimum online and offline times or CO2 emissions rate

determine among other factors in the model the units' utilization rates. This subsection presents

the assumptions made for each of the following generator's parameters:

Maximum output

Minimum technical output

Upward and downward ramp limits

Minimum online and offline times

CO2 emission rate

Maximum output

The maximum output of a generator corresponds to the capacity installed of that generator. The

capacity of every generating unit is obtained by using the ENTSO-E's Scenario Outlook &



Adequacy Forecast mentioned in section 3.1. The country's installed capacity for the

corresponding fuel type, country and scenario (scenario 2014, 2020 and the Visions for 2030) is

divided by the number of units of the corresponding fuel type and country.

The number of generators assigned to every node of the network is estimated such that the

above calculation gives installed capacity values that lie within a realistic range. The number of

units of every type of technology is assumed to be the same across scenarios.

In Appendix D, Table D.6 to Table D.11 show the capacities of the generators used for every

fuel type, country and scenario.

Minimum technical output

The minimum technical load required for every generator is assumed to be a percentage of the

nominal maximum output. Table D.12 in Appendix D indicates the adopted values according to

the type of technology taken from (Neuhoff, et al., 2013).

.Upward and downward ramping limits

Upward and downward ramp limits define by how many MW can the generator increase or

decrease its power output in one hour, respectively. Same load gradient is defined for both

upward and downward limits, see Table D.13 in Appendix D. Values were taken from Comillas

University model prototypes.

Minimum online and offline times

Flexibility of conventional generation technologies is a very important feature to measure in a

power system with an increasing share of renewable generation. However, due to both technical

and economical reasons, once a unit is connected or disconnected from the grid this one has to

remain connected or disconnected, respectively, for a minimum amount of time. Given the lack

of data for this specific parameter, the "cold" startup time from (Eurelectric, 2011) was assumed

as a good approximation, see Table D.14 in Appendix D. Both minimum online and offline

times are considered to be the same.

CO2 emission rate

Every generator unit emits certain amounts of CO2 depending on the technology used and has a

certain annual energy output. The CO2 intensity of each generator (tCO2/MWh) is the emission

rate. This parameter is calculated with the available annual CO2 emissions and the annual

energy output from Enipedia database (TUDelft, 2014). The data from Enipedia is the

projection for year 2020 and the same value is used in all scenarios.

The amount of CO2 emissions produced in the power system is included as a term of the

objective function. The aim is to minimize this term by including a high penalization cost for

emitting. This cost, also included as a model parameter, is referred to as the CO2 emission cost

rate and its value is set to 100 Euros/tonnes of CO2 taken from the (European Commission,



2013) ensuring compliance with European regulatory rules. Knowing that the chosen value is

presumably high in the range of possible actual values in the ETS trading scheme, a sensitivity

analysis will be performed with a reduced value of the base case CO2 emission cost rate and

changes in the generation costs and generation dispatch profiles will be analysed.

3.2.2.3. Initial conditions

The initial conditions of the generators such as the unit's status, its output and the initial time

status need to be fixed beforehand in order to get a realistic behaviour of the units in

consecutive periods.

Initial output

The model is run for one week plus one day in advance. This day in advance is used to specify

the initial output of every generator. The 24th hour of that day sets the initial conditions for the

actual week that wants to be modeled. These initial conditions are calculated for every scenario

defined and for comparison purposes, the three studied power market designs (Nodal, Zonal and

Copper Plate) use the same initial conditions.

Initial commitment status

Once the initial output of the units is specified, and by using Morales-España et al. (2013)

formulation, the initial commitment status is determined depending on whether the initial output

is greater or smaller than the minimum technical output of the unit.

Initial time status

This parameter indicates the time in hours that the generator has been online or offline. It is

assumed that in case the initial output of the unit is zero, the initial time status equals to minus

the minimum offline time; in other words this is the minimum time that the generator has to be

off. In case the initial output is different from zero, the initial time status equals to the minimum

online time; this is the minimum time the generator has to be switched on (Morales-España, et

al., 2013).

3.2.3. Renewable generation

Renewable energy sources are not distributed evenly over the European continent and also not

over the considered network. Solar and wind energy represent 15,7% share of renewable

generation in 2013 (Eurostat, 2015) and this share is expected to drastically increase in the

following years. Therefore, these two renewable technologies are used in the research. Solar

energy is more available in countries like France, Austria, north of Italy and south of Germany



(Figure 3.9), while there is higher potential for wind energy in northern coastal and offshore

regions of the continent (Figure 3.10).

Normalised hourly profiles from the year 2014 are used to input the renewable generation in all

scenarios. These profiles are based on the wind speed and solar irradiation estimations in every

node of the network taken from a worldwide wind and solar database for the year 2014 (Saha, et

al., 2011). These estimations are calculated based on standard turbines power curves (Wijnja-

Vlot, 2015).

The hourly renewable generation input parameter is defined as the maximum hourly wind and

solar energy produced, and since curtailment is allowed in the model, it is possible for the

algorithm to determine less renewable generation in some hourly periods.

Figure 3.9. Global irradiation (kWh/m

2) (above) and solar electricity (kWh/kWp) (below) in

European countries. Source: European Photovoltaic Technology Platform, 2006.



Figure 3.10. Distribution of wind energy density (GWh/km

2) in Europe for 2030.

Source: European Environment Agency, 2008.

With the wind and solar installed capacities from the Scenario Outlook & Adequacy Forecast

2014-2025 (ENTSO-E, 2014a), and by multiplying these capacities allocated in every node by

the normalised generation profiles, the actual generation profile is obtained in every node and

for every scenario. Both onshore and offshore installed capacities are included given that it is

estimated that a large share of the future wind capacity will be due to offshore wind farms.

Each country's wind and solar installed capacities are allocated equally among the country's

nodes. In other words, this means that in Germany's five nodes, a 20% of the country's installed

capacity for both wind and solar is assigned in every node. Table B.9 and Table B.10 in

Appendix B shows the installed capacity for wind and solar technologies for every country and

every scenario.

3.2.4. Network parameters

Transmission line capacities and line reactances are probably among the most important input

parameters in the model. These can largely affect the output of the unit commitment schedule

and power flow output. It should also be noted that there was no access to official data so

educated estimations were made in order to stay as close as possible to reality.



3.2.4.1. Transmission line capacities

Transmission line capacities are defined as the maximum power a particular line can withstand

in normal operational conditions. According to the network topology defined in Van Blijswijk,

et al., (2015), some of the transmission corridors have more than one line circuit, considering

them lines in parallel with the same power capacity.

A sensitivity analysis will be performed by reducing and increasing by 50% the capacity of the

intra-zonal transmission lines with respect to the base case values. The intra-zonal lines are the

ones within zones, i.e. countries.

3.2.4.2. Inductive reactances

As mentioned before, the inductive reactances of every line circuit becomes an important input

parameter since the Power Transfer Distribution Factors (PTDF), and hence the power flows are

determined by the relation between these reactances.

This network parameter is usually not made publicly available. Therefore, an average unitary

reactance of the Dutch power grid equal to is used as the reference

unitary reactance for the considered network (Van Blijswijk, 2011). Since the lengths of the

power lines are available, the reactances can be easily obtained. For the exact calculations of the

line reactances, the reader is referred to Appendix D.


CHAPTER 4

SIMULATION RESULTS



Exploration of the future European Electricity Market Design Chapter 4 – Simulation Results


4. Simulation Results

The following chapter presents the results obtained from the optimization model for the defined

scenarios and the studied power market designs and it deals with the main research question and

the second research sub-question. Discussions of the results are done baring in mind the

limitations of the research.

In subsequent sections the power system's variable generation costs, the degree of renewable

energy curtailed, the system's economic generation dispatch profiles and the degree of non-

served energy in the system are discussed. The extent to which the mentioned system

parameters are influenced by the amounts of renewable energies and by the design of the power

market in place is further developed here.

4.1. Power System Generation Costs

The main aim of this research is to gain a better insight on the functioning of the North-Western

and Central Europe's power system under different market designs and for different degrees of

renewable integration, as well as to obtain a rough order of magnitude of the effect of these two

dimensions on the total variable generation costs of the power system.

As a reminder to the reader, the model's objective function expresses the minimization of the

total variable generation costs of the system. Fixed costs related to new investments in

generation are out of the scope of this research. The variable generation costs include startup,

shutdown and fixed fuel consumption costs, which are fixed costs incurred due to the

scheduling of the generating units, and variable operation costs due to the generators'

production, CO2 emissions and the non-served energy.

The results show that in a high renewable scenario, more specifically for Vision 4, the total

variable generation costs of the power system when it operates under a zonal power market are

around 0,32% higher than if the power system operates under a nodal market, see Table 4.1.

While in the first phase of the zonal market (Zonal Phase I), generation costs are around 0,8%

cheaper than under a nodal market, this difference is reversed and increased to 0,32% after the

redispatch process (Zonal Redispatch Phase) is carried out by the TSOs.

A redispatch phase in a zonal market is required in order to adjust the scheduled generation in

accordance to the limits of the network. The research shows that redispatch costs in the high

renewable scenario amount to 19 Million Euros (1% of total variable generation costs) for the

whole system and for a weekly operation, while in the reference scenario these are significantly

smaller, over half the amount, 9,5 Million Euros (0,57% of total variable generation costs) (see

Table 4.1). Knowing that future extensions of network infrastructure are not accounted for in the

model, and therefore knowing that redispatch costs in Vision 4 could actually be lower due to

more cross-border interconnections helping in alleviating congestions in intra-zonal lines, the

redispatch process would still be necessary to completely fix the intra-zonal congestions.



Furthermore, the magnitude of the redispatch costs (19 Million Euros) is close to that of the

fixed generation costs, which depend in turn on the scheduling of the units. This points out the

inefficiency of the power system functioning under a zonal market. Given that the network

constraints are not jointly taken into account in the market clearing, the redispatch phase will be

inevitably necessary in the highly-meshed European network and will therefore add an extra

cost and an operational burden to TSOs.

Table 4.1. Breakdown of generation, non-served energy and redispatch costs for Scenario B

(2014) and Vision 4 (2030) for the three proposed power market designs.

Costs (Million Euros) Copper

Plate Nodal

Zonal

Phase I

Zonal

Phase I vs.

Nodal (%)

Zonal

Redispatch

Phase

Zonal

Redispatch

Phase vs.

Nodal (%)

Vision 4 (High RES scenario)

Total Variable Generation 1462,3 1882,8 1866,9 -0,8 1888,9 0,32

Fixed Generation 27,2 28,6 28,0 -2,3 28,0 -2,30

Operation 1432,3 1798,6 1783,7 -0,8 1802,7 0,23

Non-served energy 2,8 55,6 55,2 -0,8 58,2 4,72

Redispatch - - - 19,0

Scenario B 2014 (Low RES scenario)

Total Variable Generation 1559,9 1640,6 1640,5 0,0 1659,9 1,17

Fixed Generation 10,2 11,6 11,6 0,1 11,6 0,15

Operation 1549,7 1627,2 1627,0 0,0 1636,4 0,57

Non-served energy 0,1 1,9 2,0 4,7 11,9 531,66

Redispatch - - - 9,5

On the other hand, the absolute difference in total variable generation costs between nodal and

zonal markets for Vision 4 is 6,1 Million Euros (see Table 4.1). This weekly difference in costs

would translate into 317,2 Million Euros of annual savings under a nodal market. Neuhoff, et

al., (2013) estimates annual savings ranging from 0,8 Billion Euros to 2 Billion Euros under a

nodal market, depending on the penetration of renewables considered. Taking into account the

simplifications made on this thesis network compared to the one used in Neuhoff, et al., (2013),

the result obtained in this research is believed to be in line with the range estimated by Neuhoff.

With regard to the reference scenario, there is a larger difference, 1,17%, between the nodal and

zonal power market schemes with respect to the high renewable scenario. However, similarly to

before, the increase in variable generation costs can be observed in the redispatch phase of the

zonal market (see Table 4.1). Given the much lesser amount of renewable capacity available in

the system in this scenario, more expensive generating units are in charge of amplifying the

difference in relative terms between nodal and zonal in a low renewable scenario.



Additionally, both market designs have relatively similar generation costs as those that would

be expected in a copper plate market, in which one single electricity price would be established

for the whole of the European power system (see Figure 4.1). A copper plate market design is

used, as already explained, as a theoretical reference and has little real meaning, since it is based

on the unrealistic assumption that the power system has unlimited network capacity to supply

cost-efficient electricity to everywhere in the network. However, it can still be used as a

reference because it highlights the increasing value of transmission with higher renewable

energies in the system.

Furthermore, when comparing each market design between both renewable scenarios, Table 4.2

shows that total variable generation costs increase in both nodal and zonal cases in Vision 4 by

14,8% and 13,8%, respectively, while in the case of a copper plate market, costs would decrease

by 6,3%. The significant increase of generation costs in the nodal and zonal cases with large

amounts of renewable sources, together with the decrease of these in the copper plate case,

encourages to conclude that this increase might be due to a shortage in network capacity that

produces congestions and causes either more expensive technologies to be committed or

demand to not be met, and hence non-served energy costs to rise. In fact, Table 4.1 shows the

significant variations in non-served energy costs between the high and low renewable scenarios

for the zonal and nodal power market designs. However, further analysis is performed in

subsequent sections before reaching that conclusion.

Additionally, it can be noticed that the mentioned effect of an inappropriate network for a future

scenario with high renewables is amplified in the nodal market, which has a slightly higher

difference than in the zonal market (see Table 4.2). The reason for this lies in the inherent

characteristic of the nodal market to optimally utilize the generating resources in the system

across the entire system. Therefore, a poor network would distort the ideal functioning of the

nodal market scheme.

Table 4.2. Comparison of total variable generation costs between Reference scenario (2014) and

Vision 4 (2030) for the three proposed power market designs in all their phases.

Power Market Design

Reference

scenario (2014) Vision 4 (2030)

Vision 4 vs.

Reference

scenario (%) Low RES High RES

Copper Plate 1559,9 1462,3 -6,3

Nodal 1640,6 1882,8 14,8

Zonal First Phase 1640,5 1866,9 13,8

Zonal Redispatch Phase 1660,0 1888,9 13,8



Figure 4.1. Generation costs in Reference scenario B (2014) and Vision 4 (2030).



In summary, the results referred to above reveal a potential saving of 0,32% in variable

generation costs under nodal power market operation in contrast to a power system operating

under a zonal market. Especially if the large expected projections of renewable sources of

generation finally materializes, the result above could indicate that a nodal market in Europe

would allow to maximize social welfare provided that the required network capacities are

delivered effectively.

4.2. Further Analysis of RES Scenarios and Power Market Designs

The following subsections discuss some possible underlying causes of the increase in variable

generation costs in a high renewable scenario. Several output variables are analyzed such as the

curtailment of renewable generation, the total production of every type of technology and the

non-served energy in the system.

Also, in this section the electricity prices calculated a posteriori in the model are presented and

an analysis is done on how they might influence actors in the power sector and across regions in

Europe.

4.2.1 Curtailment of renewable energy

As it was mentioned in section 2.4.2, renewable energy production is given priority in the day-

ahead's merit order among the rest of generation technologies. However, it could happen that,

due to the concentration of these energy sources in the end-points of the network and a weak

transmission capacity in some specific parts of it, wind and solar power spillages not only

become necessary but could turn out to be significantly high with larger integration of

renewables. In a current zonal market total system generation costs would increase due to the

redispatch process and revenues for renewable energy producers would shrink.

The outputs of the model show that the degree of curtailment rises notably in Vision 4

compared to the reference scenario, for which there is no curtailment. Figure 4.2 presents the

distribution of the system's hourly curtailment, expressed in percentage and calculated with

respect to the maximum renewable generation. For Vision 4, it can be observed that half of the

hourly curtailment lies between approximately 1% and 9% in the nodal case and between 1%

and 9,5% in the zonal case (the interquartile range in the box-plots). Therefore, in both power

market designs the curtailment variability is very similar, with slightly higher maximum

curtailment values, above 22%, in the zonal case.

The similarity between both system curtailment profiles can be largely attributed to using the

same renewable daily patterns implicitly accounted for in the wind and solar feed-ins. The

difference between them, however, results from the required redispatch process in the zonal



market. This highlights again that provided the network capacities are limited, a nodal market

proves to make a slightly more efficient use of the available transmission capacity and promotes

a better integration of renewable energies. Nevertheless, further analysis should be made with

different time horizons and other weeks during the year, since the one modeled could have been

an exceptional one in terms of renewable generation.

Figure 4.2. Distribution of the hourly RES curtailment of the power system in Reference

scenario (2014) and Vision 4 (2030).



In absolute terms, in a power system that would generate a weekly amount close to 45,3 TWh of

energy in a extreme high renewable scenario in 2030, around 1,57 TWh of renewable energy in

the nodal case and 1,60 TWh in the zonal case would be "spilled". These amounts represent a

weekly curtailment of 7,83% and 8,01%, respectively, with respect to the weekly renewable

energy production (see Table 4.3). These curtailment values are noteworthy. A major reason

that could justify such notable amounts of curtailment in the European system is the insufficient

development of the transmission network assumed which leads to the incapability of supplying

cheap renewable energy across wide regions and having to commit or reschedule local and more

expensive technologies. In this research, no grid upgrades have been included between scenarios

for comparison reasons. However, it is clear that network planning should keep the deployment

of future renewables in mind and economic assessments should be performed to quantify the

savings made with reduced curtailment against the expensive investments of line extensions.

Table 4.3. Renewable production, total production and curtailment of renewable energy in

Reference scenario (2014) and Vision 4 (2030) for the three power market designs.

System output Copper Plate Nodal

Zonal

Redispatch

Phase

Vision 4, 2030 (High RES scenario)

RES Production (GWh) 21599,89 20031,19 19998,35

Total generation (GWh) 45591,08 45327,10 45313,98

Curtailment (GWh) 0,00 1568,71 1601,54

Curtailment (%) 0,00 7,83 8,01

Reference scenario, 2014 (Low RES scenario)

RES Production (GWh) 5646,03 5646,03 5646,03

Total generation (GWh) 36391,36 36382,58 36332,57

Curtailment (GWh) 0,00 0,00 0,00

Curtailment (%) 0,00 0,00 0,00

Besides the above, curtailment of renewables can also take place when an adequate reserves

margin is required to maintain system balance and system reliability. Figure 4.3 shows that the

time intervals in which system curtailment would be highest match with the early hours of the

day during which the highest increase of demand occurs. Before this period coal, gas and fuel

oil power plants are usually brought up to their technical minimum output and, hence, excess

night wind power output needs to be turned off. In the next section an analysis will be done with

regard to the dispatch generation profiles, and it then can be further analysed whether thermal

generation substitutes renewables in such periods.



Figure 4.3. Hourly wind and solar energy curtailment in the power system in Vision 4 (2030).

Finally, curtailment can also occur due to a global oversupply in the system. In this situation,

there is an excessive amount of generation capacity installed than what is actually needed. Such

situation can be detected by checking whether there is demand systematically not met in every

node in the network for consecutive hours. Section 4.2.3 looks further into this.

4.2.2 System Generation Dispatch profiles

This section looks into the generation dispatch profiles obtained for the system as a whole under

the three different power market designs and for the low and high renewable scenario.

The reader should be reminded that Vision 4 assumes a strong integrated European market with

harmonized policies within Member States, and this is reflected in the generation mix and in the

capacities installed in the system of each type of technology. All energy resources should be

available to meet demand everywhere else in Europe.

Compared to the reference scenario, in Vision 4 and in all three power market designs there is a

significant reduction in nuclear power usage as base load mainly to accommodate the large

amounts of renewables (see Figure 4.4). Moreover, the nuclear power generated in the high



renewable scenario follows a more variable profile suggesting greater flexibility in the operation

of nuclear power plants in the near future.

Regarding more flexible plants like coal-fired or gas-fired plants, these have undergone

substantial changes in their dispatch pattern. Both technologies become even more flexible,

experiencing drastic peaks and valleys in Vision 4. Overall, energy production from these two

technologies has experienced inverted behaviours: gas-fired plants increased their production

despite the high fuel costs in contrast to the reduction of coal-fired power plants production (see

Table 4.4). The need for quick responsive backup technologies with high penetration of

renewables and a high CO2 emission cost rate used have led to this increase in usage. With

respect to oil-fired plants, their energy output has reduced mainly to the reduction of their

capacity installed across the continent.

Renewable energy production, however, has reached very high levels in Vision 4, to the extent

that in some particular hours all the demand is covered with almost only nuclear and renewable

energy especially in the Copper Plate market (see Figure 4.4).

When comparing between power market designs, Table 4.4 shows that nodal and zonal power

markets, if yet different, are quite similar in contrast to the power outputs obtained in the copper

plate option. On the one hand, larger amounts of cheaper energy, i.e. renewables and nuclear,

are produced in the copper plate instead of using the more expensive thermal technologies. This

partly explains the cheaper variable generation costs obtained in a theoretical copper plate

system. On the other hand, in the nodal and zonal markets more production comes from gas

(18,22% and 18,46%, respectively) and coal plants (30,01% and 30,66%, respectively) due to

the limitations in the network and the redispatch process that prevent cost-efficient technologies

from being dispatched.



REFERENCE SCENARIO, 2014 VISION 4, 2030

CO

PP

ER

PL

AT

E

NO

DA

L

ZO

NA

L R

ED

ISP

AT

CH

PH

AS

E

Figure 4.4. Generation dispatch profiles in Reference scenario (2014) and Vision 4 (2030) for

the three power market designs




and Vision 4 (2030) for the three power market designs.

Production by

technology

type (GWh)

Copper

Plate Nodal

Zonal

Redispatch

Phase

of Nodal

with respect to

Copper Plate

(%)

of Zonal

with respect to

Copper Plate

(%)


Gas 7528,04 8899,65 8917,84 18,22 18,46

Coal 4040,22 5252,62 5279,10 30,01 30,66

Nuclear 11968,85 10594,33 10557,51 -11,48 -11,79

Oil 454,07 549,32 561,18 20,98 23,59

RES 21599,89 20031,19 19998,35 -7,26 -7,41

Total 45591,08 45327,10 45313,98 -0,58 -0,61


Gas 5344,37 5769,47 5536,28 7,95 3,59

Coal 6246,18 6580,41 6970,96 5,35 11,60

Nuclear 17503,77 16996,99 16754,54 -2,90 -4,28

Oil 1651,02 1389,68 1424,76 -15,83 -13,70

RES 5646,03 5646,03 5646,03 0,00 0,00

Total 36391,36 36382,58 36332,57 -0,02 -0,16

If a further step is reached in the analysis and due attention is given to what is happening in the

system's operating phases under a zonal power market, the reasoning by Neuhoff et al. (2011) is

proved: zonal market produces an inefficient dispatch within countries.

The reason is again network limitations. After the first phase, a commitment schedule ignoring

intra-zonal network limits, i.e. congestion, that ensures minimum system generation costs is

obtained. The second phase takes care of the infeasibilities in the schedule caused by line

congestions. This means that cheap generation that was aimed to supply a load center located

somewhere in the network can no longer fulfill this due to the impossibility of being delivered.

Instead, local and more expensive generation covers that shortage and the cheap generation i.e.

renewable and nuclear energies face curtailments. As Table 4.5 shows, around 1,82% more gas

and 1,82% more oil are dispatched in a high renewable scenario compared to the first phase.

Similarly, nuclear and renewable plants are redispatched and their production decrease by

0,88% and 0,39%, respectively.

It is noteworthy to mention, however, that compared to a low renewable scenario the redispatch

of expensive generation sources is reduced in relative terms, mainly due to the greater overall

availability of renewable capacity installed in the system.




and Vision 4 (2030) for zonal power market in its two phases.

Production by

technology type

(GWh)

First Phase (I) Redispatch

Phase (II)

of Redispatch

phase with respect to

First Phase (%)


Gas 8758,23 8917,84 1,82

Coal 5292,10 5279,10 -0,25

Nuclear 10650,72 10557,51 -0,88

Oil 551,16 561,18 1,82

RES 20077,06 19998,35 -0,39

Total 45329,28 45313,98 -0,03


Gas 5276,78 5536,28 4,92

Coal 6962,33 6970,96 0,12

Nuclear 17132,04 16754,54 -2,20

Oil 1368,64 1424,76 4,10

RES 5646,03 5646,03 0,00

Total 36385,82 36332,57 -0,15

On the one hand, these results reveal that managing national congestion during the operational

hour, after market clearance i.e. after a unit commitment schedule is generated, rather than

through the market, as it is done under a nodal scheme, it ultimately produces higher generation

costs especially when the network transmission capacities are designed so that small levels of

congestion are always produced. On the other hand, the increased amount in renewable

generation in the high RES scenario produces variations in the merit order dispatch with respect

to that of the reference scenario, which in turn highlights the changes undergone with each

market design as well as the slight differences between them.

4.2.3 Non-served energy

This section analyses to what extent the non-served energy costs impact on the system's

dispatch and on the system's variable generation costs in the different power market designs and

in the high and low renewable scenarios.

As previously stated in section 4.1, the magnitude of the non-served energy costs are

considerably high for the nodal and zonal markets in a scenario with high penetration of

renewable energies. Costs of unserved energy in the zonal market amount to 58,22 Million

Euros, a 4,72% larger than those in the nodal market, 55,60 Million Euros (see Table 4.6). In the

reference scenario, however, non-served energy costs only amount to 1,88 Million Euros in the



nodal market and to 11,9 Million Euros in the zonal market. The latter one is still considerably

high.

Table 4.6. Comparison of non-served energy costs between Reference scenario (2014) and

Vision 4 (2030) for the three proposed power market designs in all their phases.

Costs (in Million Euros) Copper

Plate Nodal

Zonal

Phase I

Zonal

Phase I vs.

Nodal (%)

Zonal

Redispatch

Phase

Zonal

Redispatch

Phase vs.

Nodal (%)

Vision 4 (High RES scenario)


Non-served energy 2,80 55,60 55,16 -0,78 58,22 4,72

Non-served energy : Total

Variable Generation (%) 0,192 2,953 2,955 3,082

Scenario B 2014 (Low RES scenario)


Non-served energy 0,12 1,88 1,96 4,74 11,9 531,66

Non-served energy : Total

Variable Generation (%) 0,008 0,114 0,120 0,717

These costs largely depend on the system's generation adequacy to meet the required demand,

reflected by the costs in a Copper Plate market in Table 4.6, and the system's network capacity

planning to appropriately deliver the energy to all demand points in the network. Non-served

energy costs of nodal and zonal markets represent 2,95% and 3,08% of the total system's

variable generation costs, respectively (see Table 4.6). It is again highlighted the importance of

a timely delivery of the network infrastructure investments in order to gradually integrate the

large deployment of renewables in the system.

4.2.4 Effects of system outputs on the generation costs

In the subsections above, it has been discussed the influence of a large penetration of renewable

energies in the power system and the design of the latter on the following system outputs:

curtailment of renewable energy, system's economic generation dispatch profile and non-served

energy in the system.

As a summary, Table 4.7 shows the contribution of the system outputs on the system's variable

generation costs. It should be noted that while the energy production by type of technology and

the non-served energy directly impact on the costs of generation in the system, the curtailment

of renewable energy does not directly contribute to it. The degree of curtailment of renewables



instead shows the lost savings in the system's generation dispatch, the opportunity cost of

having to spill renewable generation because of not having the sufficient physical network to

deliver that green cheap energy.

Table 4.7. Impact of system outputs on variable generation costs.

System output

(GWh)

Copper

Plate Nodal

Zonal

Redispatch

Phase

of Nodal

with respect to

Copper Plate

(%)

of Zonal

with respect

to Copper

Plate (%)


Gas 7528,04 8899,65 8917,84 18,22 18,46

Coal 4040,22 5252,62 5279,10 30,01 30,66

Nuclear 11968,85 10594,33 10557,51 -11,48 -11,79

Oil 454,07 549,32 561,18 20,98 23,59

RES 21599,89 20031,19 19998,35 -7,26 -7,41

Non-served energy 14,01 277,99 291,11 1884,23 1977,82

Curtailment 0,00 1568,71 1601,54 Very large Very large


Gas 5344,37 5769,47 5536,28 7,95 3,59

Coal 6246,18 6580,41 6970,96 5,35 11,60

Nuclear 17503,77 16996,99 16754,54 -2,90 -4,28

Oil 1651,02 1389,68 1424,76 -15,83 -13,70

RES 5646,03 5646,03 5646,03 0,00 0,00

Non-served energy 0,60 9,38 59,39 1461,51 9785,13

Curtailment 0,00 0,00 0,00 0,00 0,00

Overall, a combination of factors leads to having higher variable generation costs in a scenario

with high penetration of renewable technologies, as presented in section 4.1. As indicated in

Table 4.7, a drastic increase of renewable production would contribute to a significant decrease

in generation costs but a significant increase in gas production to support the unpredictability of

renewables and maintain the system's reliability within adequate limits has the opposite impact.

Simultaneously, around a 60% decrease in cheap nuclear production accentuates this effect even

more. Furthermore, the high renewable scenario results in very large amounts of non-served

energy due mainly to insufficient network capacity and large amounts of curtailment due again

to congestions, i.e. network inadequacy, that prevent cheap renewable energy to be transported

to load centers far away.

With respect to the power market designs, and taking Copper Plate as a reference, the nodal and

zonal markets experiment higher amounts of gas, coal and oil, i.e. more expensive technologies,

and a reduction in nuclear and renewable energy productions. As for non-served energy and



curtailment levels, these are extremely above those of the Copper Plate market (see Table 4.7).

Additionally, it should be mentioned that, in a high renewable scenario, nodal and zonal markets

have very close percentage differences between them for every system output, when compared

to Copper Plate. Finally, and referring back to the second research sub-question, it all comes

down to having an adequate planning to expand the network and be able to follow up a

complete integration of electricity markets and the deployment of the renewable targets.

Therefore, whether it makes operational and economical sense to implement a nodal market in

Europe will depend on the network expansion projects and the success of cooperation policies

between Member States.

4.2.5 Electricity Prices and Congestion Costs

This section presents an overview of the electricity prices calculated following the model's

economic dispatch outcome. More specifically, the weekly average electricity prices in the

nodal and zonal markets and the difference between them are presented for both high and low

renewable scenarios.

Wholesale electricity prices under a nodal market scheme bring many short-term benefits as

already identified in previous sections. They provide more transparency to market players, more

efficiency and reliability in system operation for TSOs at the same time they reduce local

market power practices and unwanted trading strategies, the so-called "inc-dec game" (Neuhoff,

et al., 2011) and support the transmission planning process. Moreover, demand responsiveness

would also benefit from its implementation in retail markets.

Figure 4.5 indeed reveals an overall decreasing trend in weekly average nodal prices in a higher

renewable scenario in both market designs. Furthermore, in the case of the nodal market both

intra-national and cross-border network congestions can be observed in the reference scenario

between some regions: between North and South of Germany, between Czech Republic and

Germany, between France and Germany, between Belgium and France or within Belgium. This

suggests the need for transmission capacity development both within national power grids and

between international borders. Also, these congestions, especially in continental Europe, seem

to be alleviated with the presence of more renewables. Most probably this is due to the higher

amounts of renewable generation that meet the demand in every node of the network.

Nonetheless, congestions keep existing between the end points of the network and continental

Europe.

In the case of the zonal market, the nodal prices reflect the single electricity price per zone, i.e.

per country, and as it can be observed in Figure 4.5 there is also a tendency towards greater

homogenization of prices with larger amounts of renewable energies, especially in central-

western Europe. The extremes of the network, however, have either much cheaper electricity

prices due probably to a higher concentration of renewable sources or much more expensive

prices. The latter could be due to either a lack of transmission capacity or a still quite

predominant thermal generation mix, like the case of Poland, in which prices are pushed

upwards through the merit order mechanism.



REFERENCE SCENARIO, 2014

VISION 4, 2030

NO

DA

L

ZO

NA

L F

IRS

T P

HA

SE

Figure 4.5. Average Locational Marginal Prices in nodal and zonal markets in Reference

scenario (2014) and Vision 4 (2030).

Through the operational information given in real-time by TSOs, including the information

about the state of the network provided by the LMPs, generators can identify when and where

congestions and supply shortages are taking place allowing them to consider it in their short-

term decision-making. From the consumer's point of view, however, only those that explicitly

bid into the wholesale market are directly impacted by the prices. The rest of consumers who



are supplied by electric utilities do not experiment the full impact of the LMPs and their

variations for now, since utilities average prices across the territories where they own generating

units. Only competitive retail markets that facilitate demand response initiatives would give the

opportunity to consumers to react to price spikes.

In order to check for differences in LMPs between power market designs and how these might

affect participants in the European power market, Figure 4.6 plots the difference between

weekly average nodal prices under both market designs, calculated as:

.

Overall, the differences between nodal prices, if already small, are generally even smaller in a

higher renewable scenario, as shown in Figure 4.6. Moreover, energy producers get paid at the

node where their generators are located. Therefore generators located in zones like the south of

Germany, Netherlands, Austria and north of Italy, for instance, where the differences in LMPs

are negative, would benefit from the implementation of a nodal market scheme since LMPs in

nodal would be higher than those in a zonal market. Figure 4.6, in fact, illustrates that

implementing a nodal market in Europe would impact to a greater or lesser extent on the

revenues of the energy producers across the continent, both within national borders and outside

a country's borders. This result suggests that a harmonization of the energy policies related to

capacity markets is necessary in the different Member States to minimize welfare redistributions

issues.

However, it should be taken into account, though, that the modeled week could be an

anomalous one in terms of renewable generation for example. Therefore, a more in-depth

analysis of the electricity prices is needed to conclude the degree of impact of the market design

and the large amounts of intermittent generation on the LMPs and therefore on the market

actors. The price analysis should be done for a longer time horizon and in different seasons of

the year.



REFERENCE SCENARIO, 2014

VISION 4, 2030


in Reference scenario (2014) and Vision 4 (2030).

4.2.6 Sensitivity analysis of model parameters

In the following subsections a sensitivity analysis is performed to test the impact of the CO2

emission cost rate and the intra-zonal network capacities on the outcomes of the model, mainly

on the system variable generation costs, the economic dispatch profiles and the nodal prices

under the nodal and zonal market designs. The sensitivity analysis is performed for the high

renewable scenario and it is contrasted against that of the base case.

4.2.6.1. CO2 emission cost rate

The base case value for the CO2 emission cost rate was set to 100€/tonnes of CO2, which is

thought to be on the high side of the forecasting range. Therefore, a significantly lower value of

20€/ tonnes of CO2 is chosen for the sensitivity analysis.

As Table 4.8 shows, with a CO2 cost five times lower, total variable generation costs in a high

renewable scenario would be 0,58% higher in a zonal market with respect to a nodal market, in

contrast to the 0,32% difference in the base case (see Table 4.1). Besides, operation costs are



lower as cheaper coal generation is dispatched instead of expensive gas units. Figure 4.7 indeed

shows that part of the base case gas production has been replaced by production of coal.


for the nodal and zonal power markets and for the base case and low CO2 emission cost rate.

Costs (Million

Euros) Nodal

Zonal

Phase I

Zonal

Phase I vs.

Nodal (%)

Zonal

Redispatch

Phase

Zonal Redispatch

Phase vs. Nodal

(%)

Low CO2 emission cost rate = 20 €/tonnes of CO2

Total Variable

Generation 1106,9 1090,2 -1,5 1113,3 0,58

Fixed Generation 28,4 29,4 3,5 29,4 3,50

Operation 1055,7 1038,8 -1,6 1054,9 -0,08

Non-served energy 22,7 22,0 -3,1 29,0 27,69

Redispatch - - 16,1

Base case

Total Variable

Generation 1882,8 1866,9 -0,8 1888,9 0,32

Fixed Generation 28,6 28,0 -2,3 28,0 -2,30

Operation 1798,6 1783,7 -0,8 1802,7 0,23


Redispatch - - 19,0



BASE CASE

LOW CO2 EMISSION COST RATE

NO

DA

L

ZO

NA

L R

ED

ISP

AT

CH

PH

AS

E

Figure 4.7. Generation dispatch profiles in Vision 4 (2030) for the nodal and zonal power

market designs for the base case and low CO2 emission cost rate.

With regard to the nodal prices of electricity, while a lower CO2 emission cost does not

influence to a large degree, in principle, the network's congestion patterns, Figure 4.8 does show



reduced nodal prices in both market designs with a lower CO2 price. From the policy

perspective, this underlines the importance of setting appropriate tax levels or subsidies that

could otherwise distort the market outcome to undesirable levels.

BASE CASE


NO

DA

L

ZO

NA

L F

IRS

T P

HA

SE


markets for the base case and low CO2 emission cost rate.

A lower CO2 price also makes the difference between the electricity prices in the zonal and

nodal markets more negative across the network (see Figure 4.9), thus, favouring the nodal

market in the generators' view as revenues would be larger. However, some zones in northern



Germany and Denmark would still slightly benefit from a zonal market. Again, a broader price

analysis should be performed to be able to conclude more firmly.

BASE CASE



in Vision 4 (2030) for the base case and low CO2 emission cost rate.

4.2.6.2. Intra-zonal network capacities

Transmission lines and network capacities represent important factors in the modeling of the

power system since they determine the power flow profiles and the power limits of the lines.

Two sensitivity tests will be performed to analyse the impact of doubling or halving intra-zonal

capacities on system variable generation costs, on nodal prices and on the difference of nodal

prices between nodal and zonal markets. The intra-zonal network capacities are the line

capacities located within a country.

4.2.6.2.1. Double intra-zonal network capacities

Increasing by 50% the capacities of the intra-zonal transmission lines would reduce the

difference in total variable generation costs between the nodal and zonal markets to 0,17% as

compared to the 0,32% of the base case (see Table 4.9). However, a 4,38% difference in non-

served energy costs between zonal and nodal markets is still close to that of the base case,

4,72%. Plus, absolute values of non-served energy costs are very similar to those in the base

case. Moreover, total variable generation costs would be reduced by only 0,5% and 0,65%, in

the nodal and zonal markets respectively, compared to the base case. This suggests that



doubling the capacities within zones would not be enough to meet system's demand in a cost-

efficient way.


for the nodal and zonal power markets and for the base case and double intra-zonal capacities.

Costs (Million Euros) Nodal Zonal

Phase I

Zonal

Phase I vs.

Nodal (%)

Zonal

Redispatch

Phase

Zonal

Redispatch

Phase vs.

Nodal (%)

Double intra-zonal network capacities

Total Variable Generation 1873,4 1866,9 -0,4 1876,6 0,17


Operation 1789,1 1783,7 -0,3 1790,9 0,10


Redispatch - - 7,1

Base case



Operation 1798,6 1783,7 -0,8 1802,7 0,23


Redispatch - - 19,0

In terms of alleviating network congestions, doubling the capacities within countries has slightly

helped to improve transmission constraints in the nodal market between the north and south of

Germany and within Belgium, as Figure 4.10 shows. Additionally, prices in the nodal market

are very similar to those in the zonal market. Nevertheless, cross-border congestions limit the

nodal market's potential to improve operating efficiency of the system since they keep being

quite significant.



BASE CASE

DOUBLE INTRA-ZONAL CAPACITIES

NO

DA

L

ZO

NA

L F

IRS

T P

HA

SE


markets for the base case and double intra-zonal capacities.

On the other hand, a 50% increase in the intra-zonal network capacities decrease the difference

in nodal prices between both markets, compared to the base case, to values that range from -4 to

2 €/MWh according to Figure 4.11. From the generators' point of view, the small differences

would hardly impact on their revenues if a nodal market is implemented. However, a more

extensive study should be done in order to account for possible anomalies in this week's

generation profile.



BASE CASE

DOUBLE INTRA-ZONAL CAPACITIES


in Vision 4 (2030) for the base case and double intra-zonal capacities.

4.2.6.2.2. Half intra-zonal network capacities

In the case of decreasing by 50% the intra-zonal capacities, the zonal market turns out to be

1,89% more costly in terms of total variable generation costs than the nodal market design. In

comparison to the base case, total variable generation costs would be 3,36% higher in the nodal

market and almost 5% higher in the zonal market. Moreover, in this case the redispatch costs

would amount to 53,6 Million Euros, 2,82 times higher than in the base case, and non-served

energy costs in the zonal redispatch phase would more than double those in the first phase,

reaching 117,7 Million Euros (see Table 4.10).



Table 4.10. Breakdown of generation, non-served energy and redispatch costs in Vision 4

(2030) for the nodal and zonal power markets and for the base case and half intra-zonal

capacities.

Costs (Million Euros) Nodal Zonal

Phase I

Zonal

Phase I vs.

Nodal (%)

Zonal

Redispatch

Phase

Zonal

Redispatch

Phase vs. Nodal

(%)

Half intra-zonal network capacities



Operation 1860,2 1783,7 -4,1 1837,4 -1,23


Redispatch - - 53,6

Base case



Operation 1798,6 1783,7 -0,8 1802,7 0,23


Redispatch - - 19,0

In relation to the electricity prices, reducing intra-zonal line capacities accentuates network

congestions within zones even more. This can be observed in north-eastern Germany and

Belgium in the nodal market, whereas in the zonal case no major change can be seen (Figure

4.12). Cross-border transmission constraints are again not affected. Here, it is reflected the

transparency of the nodal market over the zonal one. The nodal market provides information

about where transmission development would be more valuable to reduce congestions in the

network while the zonal approach conceals it completely.



BASE CASE

HALF INTRA-ZONAL CAPACITIES

NO

DA

L

ZO

NA

L F

IRS

T P

HA

SE


markets for the base case and half intra-zonal capacities.

Once again, in this case a 50% decrease in intra-zonal line capacities increase the differences

between nodal prices of both markets to values ranging from -10 to 25 €/MWh (Figure 4.13).

These magnitudes, yet low, could start to impact producers' behaviours in the market. Energy

producers owning generators in nodes with a positive difference, meaning

, would rather operate under a zonal market structure, especially if those generators

are downstream of a transmission constraint since that would mean that once the congestion is

solved the price at that particular node will rise, hence increasing the generators' revenues.



BASE CASE

HALF INTRA-ZONAL CAPACITIES


in Vision 4 (2030) for the base case and half intra-zonal capacities.


CHAPTER 5

CONCLUSIONS AND

RECOMMENDATIONS



Exploration of the future European Electricity Market Design Chapter 5 – Conclusions and Recommendations


5. Conclusions and Recommendations

In this last chapter the main results and conclusions drawn from the research work are

presented. Section 5.1 formulates the answers to the main research question introduced in

chapter 1 as well as to its three sub-questions. Section 5.2 reflects on the modeling process with

the research work. Section 5.3 gives some recommendations for future research and relevant

considerations for policy makers in the European power sector. Some final reflections on these

policies are also exposed.

5.1. Conclusions of the research

This research work started with the objective of gaining better insights on the functioning of the

European power sector and market under different power market designs and in high renewable

scenarios. In particular, the main research question posed is:

To what extent can the power market design and the future expected large-scale integration of

renewable energy sources have an effect on the system's variable generation costs of

electricity?

High penetration of renewable energy sources bring a number of opportunities to many actors in

the power system but involve also many big challenges to power system operators. Given their

unpredictable nature and the need of a reliable and continuous system balance, the dispatch of

large amounts of this green technology in combination with other existing conventional

technologies will unfold new ways of operating the power market, starting by identifying the

future transmission line congestions in the network. A power market design able to intrinsically

manage congestion not only will play a positive effect on the integration of renewable energy

sources but will also ensure the operation of the system at minimum costs.

The current power market rules in Europe respond to a multi-region day-ahead market coupling

design (zonal market) in which real-time markets take into account inter-zonal congestion but

establish a single electricity price inside each zone, namely a country. On a second phase, this is

during operational hour, network congestion is handled by the Transmission System Operator

(TSO) who conveniently redispatches generating units, incurring in extra operational costs. In a

locational marginal pricing market scheme (nodal market), however, real-time markets

determine the feasible generating schedule at one go and electricity prices internalize the

locational value of delivering the energy, providing actors with a more transparent and efficient

market design. The implementation of a nodal market design in Europe could lead to a potential



saving of 0,32% in the system's total variable generation costs with respect to the zonal power

market option.

A successful implementation of the power market's redesign also depends on an appropriate

network expansion plan. The research has shown that a well-developed network infrastructure is

a crucial factor in the effective integration of large amounts of renewable energies. The

mentioned expansion plan should cover both already existing transmission lines that need

reinforcements within and between regions and new ones that will become necessary when

more off-shore wind farms are developed. An extended network could emphasize to a greater

extent the 0,32% savings in system's generation costs under a nodal market compared to those

in a zonal market.

Sub-question 1. How can the European power system be modeled to best reproduce its

behaviour when functioning under different power market designs in combination with high and

low renewable scenarios?

Firstly, and referring to the main research question, the objective of power systems is to supply

the required demand with the available energy resources in the market at minimum cost. For

this, TSOs use unit commitment models and constrained economic dispatch models to obtain a

generating schedule that involves minimum total generation costs for the system and that is

technically feasible with the physical network. Accordingly, the chosen technique to model

Europe's power system is based on the optimization of energy resources from a system's

viewpoint, i.e. minimization of the system's variable generation costs.

Secondly, in the context of the Internal Energy Market for Europe that aims for a competitive

and efficient power system, it was chosen to model the power system from a centralised point of

view as if there was a single centralized planner managing all the generating units in Europe. In

this way, it is assumed that the system's optimal solution is found.

Thirdly, a major feature of power market designs in relation to the penetration of renewable

energies in the system concerns the network capacity allocation method and the mechanism

used to manage congestions in the lines. While in a zonal market congestion is handled through

system operators with a redispatch process during the operational hour, the nodal market

implicitly takes care of congestion through the market clearing by including all the network

constraints in the optimization algorithm. It is precisely this difference that is used to reproduce

and model both power market designs. This method reflects the efficient use of the available

transmission capacity in the nodal market and the impact of the redispatch process on the

system's variable generation costs in the zonal market. Moreover, the additional features

included in the optimization's problem formulation related to the HVDC transmission lines in

the system give a deep insight on the complexity of the network and the system's power flow

problem.



Finally, the use of markedly different renewable scenarios validated by the ENTSO-E to model

the system's generation mix aids to gain a clear insight, from the technological point of view, of

how the different power plant technologies and their production levels in the day-ahead market

are impacted by the network's topology limitations and the power market design in question.

Sub-question 2. To what extent the difference in variable electricity generation costs between

scenarios and across power market designs can be attributed to an increase usage of renewable

technologies, network congestions or non-served energy costs?

The research has showed that the large-scale integration of renewable energies in the power

system by 2030 would lead to a reduction of 6,3% in the considered system's total variable

generation costs if the latter would operate under a Copper Plate design. A lower percentage of

savings would be expected provided the completion of the mechanisms to fully integrate the

day-ahead markets and the completion of the required network developments that prevents

significant amounts of network congestions.

Regarding the other two power market designs, despite the large rise of renewable energy

production in 2030, accounting for 44,19% and 44,13% of the total energy production in the

nodal and zonal markets, respectively, the power system has also required to increase the share

of gas production from 15,9% to 19,6% in the nodal market and from 15,2% to 19,7% in the

zonal market with respect to the reference scenario. Additionally, the incapability of the system

to meet significant amounts of energy has been reflected in the share of non-served energy costs

with respect to the system's total variable generation costs, 2,95% and 3,08% shares in the nodal

and zonal markets, respectively. Moreover, significant renewable curtailment values of 7,83%

and 8,01% in the nodal and zonal markets, respectively, reveal the system's inefficient

integration of the deployed RES capacity in the system. And last but not least, the additional

redispatch costs incurred in the zonal market, a 1% share of the total variable generation costs,

stress even more the degree of intra-zonal congestions needed to be fixed compared to the nodal

market.

All the factors above have revealed the extent of the network's congestion that provokes a share

of the demand to not be met, part of the wind and solar potential to be curtailed and

consequently more expensive technology to be dispatched, thus undermining the performance

of an efficient power market design. The above emphasizes the relevance of having an upgraded

network, whose expansion plan should be rigorously developed at the same pace as the

deployment of the projected renewable energies. In order to be successful this process needs a

common approach between Member States based on information exchanges and maximum

transparency as well as full commitment to the project of the IEM for Europe.



Sub-question 3. How can the difference in electricity prices within and between European

countries impact the markets' participants behaviour in the power market?

For a given market design, variations of electricity prices within and between countries impact

on the generators' revenues depending on whether they are located upstream or downstream of

the transmission constraint. On the one hand, generators' revenues are directly proportional to

the nodal prices of the nodes where they are located. On the other hand, being located upstream

of the transmission constraint translates into initially having assigned a higher electricity price

than that of the downstream end. In the case of the zonal market, once the congestion is fixed by

the TSO, the price of the upstream node would be lowered and the price of the downstream

node would be increased. Therefore, generators located downstream of the transmission

constraint would see an opportunity to increase their revenues. It should be noted that the more

open the market is to competition and the lower the concentration of the wholesale market is,

the less opportunity there is for incumbent generators to strategically bid in the market, the "inc-

dec game" (Neuhoff, et al., 2011).

The research shows that in both nodal and zonal market designs weekly average nodal prices

decrease in overall in a high renewable scenario. Additionally, intra-national and cross-border

network congestions can be spotted, but with the analysis performed it is difficult to conclude

whether these congestions are systematically present in the network. Also, these spotted

congestions seem to be slightly reduced with more renewables. This could be due to the overall

larger amount of renewable generation available in every node to meet the demand in each

node. Following this, a point could be reached in which, provided network capacities are

adequately developed, the vast presence of renewable energies in the generation mix could push

wholesale prices to such low levels that revenues of both thermal and renewable energy

producers could be negatively affected.

Furthermore, the base case price analysis performed shows that indeed there are some zones in

the network, and hence generators located there too, that would benefit from the implementation

of a nodal market since nodal prices would be slightly higher under a nodal scheme. However,

for the modeled week these price differences between market designs are reduced as more

renewables are integrated. Taking into account the shortcomings of the network modeled, higher

differences between nodal and zonal nodal prices could be expected.

On the other hand, the results of the sensitivity analysis reflect that the lower the price that

thermal generators have to pay for CO2 emissions, the more negative are the nodal price

differences between the zonal and nodal markets, and hence the more economically beneficial

would be for generators bidding in the market. This output reminds that any kind of

environmental policy or subsidy can impact, to a larger extent than thought, the market outcome

and therefore they should be looked at carefully before implementing them. Moreover, a 50%

increase in intra-zonal network capacities in a high renewable scenario would decrease the

difference in nodal prices between both markets, to significantly smaller differences that would

not largely impact on producers' revenues. In contrast, a 50% decrease in intra-zonal line

capacities would increase nodal price differences to values of up to 25€/MWh. These



magnitudes, if prolonged in time, could start impacting market participants and define whether a

zonal or nodal market would be more convenient from the producers' perspective.

The price analysis performed for a future high renewable scenario, if extended to longer time

horizons and different seasons, could provide further insights and a robust analysis of future

congestion patterns and a possible configuration of smaller price zones for continental Europe to

operate in a similar way to how the Nordic countries currently operate.

5.2. Reflection on the modeling process and results

Following the conclusions and the answers to the research questions posed in this thesis work, a

reflection is presented below on the assumptions and the modeling process, the results and the

limitations of the work.

As in every model representing a real-life system, many simplifying assumptions have to be

made in order to model the system under computational and time constraints. In other cases,

input data assumptions are needed due to the lack of accessible real data. In this regard, day-

ahead markets of three phase electric power systems are usually modeled using unit

commitment and transmission constrained economic dispatch models that use a linearized

mathematical formulation, the so-called DC power flow model. As explained in section 2.2, the

accuracy of the DC approximation to model the AC power network lies around 5% when it is

applied to high voltage grids and when outcomes are averaged over all the lines in the network.

This means that error deviations are acceptable as long as conclusions are not drawn for

individual lines.

In relation to the required input data, it is worth mentioning that an extensive data search was

necessary to comply with data as realistic as possible and many intermediate calculations with

this data were required to prepare the input parameters before running the optimization model.

Data assumptions related to the network are most probably among the most determining

parameters to obtain power flows that conform more closely to reality. Transmission line

capacities, the number of circuits in every transmission corridor and especially the relationship

between line inductive reactances, are key in determining these power flows since the Power

Transfer Distribution Factors (PTDFs) used in the DC power flow approximation are based on

relations between reactances. Electrical characteristics of the power grids are usually not

accessible to the public, therefore a reference impedance base was estimated (Appendix D) and

relations between reactances of the lines were kept by using the lengths of the transmission

lines.

The optimization problem is modeled with deterministic renewable scenarios that are based on

scenario data provided by the ENTSO-E. Hydro energy is, however, left out from the generation

mix due to lack of data. In order to account for this, the demand parameter was adjusted in those

nodes of the network where hydro resources are predominant, such as in Norway or Austria.

The outputs obtained from the model should therefore be interpreted knowing that optimization



of hydro resources in combination with intermittent renewable sources will play an important

role important role in the replacement of expensive thermal generation thanks to their flexibility

and complementary use. Moreover, renewable generation of wind and solar is assigned to every

node of the network according to the forecasted wind speed and solar radiation time-series in

the assumed location of that node rather than on the average of a region covered by that node.

This assumption induces to errors due to regional variations of the renewable sources. In the

same way, the initial plan was to model the power system for two different snapshots in a year,

i.e. two weeks in different seasons, in order to account for seasonal characteristic variations in

wind and solar generation profiles. However, this was not possible due to lack of time and the

many difficulties encountered to correct the model's inconsistent outputs.

It is important to remind the reader that the purpose of the thesis is not to predict exact values of

certain model outputs but rather the interest lies in revealing trends and mechanisms that could

serve for the future European power market and its design.

All in all, it should be pointed out that the outcomes obtained from the model should be

interpreted baring in mind the shortcomings of the modeled network, which despite the

similarity to reality of the cross-border interconnections, transmission lines within the countries

are significantly simplified and therefore constraints in these parts of the network are not

realistically taken into account. It is expected that with an extended and validated network

topology, the outcomes of the model regarding total variable generation costs, renewable

curtailment levels, non-served energy levels and the differences between nodal prices would

emphasize greater differences between the nodal and zonal markets, favouring even more the

nodal market scheme as an efficient alternative market design.

5.3. Recommendations

5.3.1. Guidelines for future work

Several recommendations for future research can be drawn from the present research work.

While the interaction between conventional generation and different integration levels of

renewable energies has been tackled in this thesis, the combination and complementary use of

hydro energy resources with intermittent and uncertain energy sources like wind and solar has

not been included. Hydro resources are also expected to increase, yet in a more moderate way

than pure renewable energies, and their known flexibility to stabilize variations between

demand and supply could play a major role in the future power system. In the context of

Europe's redesign of the power system it would be interesting to study how different the market

outcomes would be, especially in terms of system generation costs and network congestions.

Another aspect that could be reviewed is the perfect competition behaviour assumed in the

modeling of the system. Now, it is assumed that the generators' bids reflect their marginal costs,

however, in reality there is always strategic bidding to some extent. This ultimately impacts on



the optimal economic dispatch and the wholesale electricity prices. However, introducing

strategic behaviour of the market agents increases the complexity of the model even more.

Regarding the model's temporal scope, it is suggested that a longer time horizon is chosen to

model the power system in order to obtain more robust results and conclusions. The chosen

week in July could have been an unusual one in terms of renewable generation. For example,

different weeks during the year or more consecutive weeks would be equally acceptable

approximations.

From the network topology point of view, an increased granularity of the network to obtain a

highly-meshed network closer to the real one could bring some added value to further help

differentiate between the market outcomes of a nodal market and a zonal one. It would be

helpful to avoid end-nodes in the network to minimize congestions that in reality would not

have been there.

Also related to the network and in a similar way as was discussed in this thesis for the high-

voltage power grid, changes to how congestion is managed in the distribution system will be

required. A significant share of the projected renewable sources are expected to be integrated

through the medium-voltage distribution systems. This will cause bidirectional flows, which

were inexistent before, and they need to be taken into account. Therefore, more interaction

between the Distribution System Operators (DSO) and the market will be expected. A similar

analysis to this work could be done to study the impact of high penetration of renewables in the

efficient operation of distribution networks.

Given the further aim of integrating short-term markets and due to increasing amounts of

renewables in the system, these must increasingly be able to provide balancing power. To date,

balancing reserves have been supplied by large conventional power plants and pumped-storage

facilities. Some studies have already proved that renewable energies can also provide balancing

power, even reacting faster than other conventional plants. However, the current framework of

balancing markets prevents renewable energies from actually offering this service. An analysis

on the impact of large penetration of renewables on the design and on balancing market prices

could bring some light to this issue.

5.3.2. Reflection and policy implications in the European framework

The redesign of the power market involves many concerns for market players who in some way

or another may be affected if implementation steps go ahead. In this section I display several

reflections on some policy implications and recommendations that should be borne in mind in

parallel to the redesign process of the power market.

The redesign process of Europe's power market design attempts to achieve a single and greater

competitive and efficient power system. However, while market coupling has proved to

improve market outcomes, regional regulations can have a negative effect on the performance of

this coupled market. The national renewable policies set could tend to fragment the common

market and undermine the benefits from increased energy trade. This is because renewable



support schemes are still defined on a national basis and in many cases these schemes support

the development of a particular technology in locations where, if looked at it from a European

point of view, they are not as efficient as they would have been if placed somewhere else in the

continent. An example of this could be the deployment of large amounts of solar PV panels in

Germany compared to the not as many and as effective as those in Spain. This misallocation of

resources, along with the excessive investments and corresponding subsidies, could have been

avoided if a European framework for renewable energies would have been in place at the same

time the path for the IEM was being prepared.

Harmonizing renewable policies in all Member States and replacing feed-in tariffs on a national

level for a quota system on a European level could help promote cost-effective technologies in a

technology-neutral way and in efficient locations. A quota system is already set up in some

Nordic countries. This support scheme is quantity-based rather than setting a fixed price for

renewable producers. The renewable energy quota refers to the minimum annual share of RES

that electricity suppliers, large electricity consumers or electricity retailers have to meet and the

high compatibility with market competitiveness and the hints it gives to transmission planners to

adjust to network development requirements makes this support scheme a good candidate to

apply on a European level.

A similar issue to the above could happen in relation to securing the adequacy of national

capacity and reserve mechanisms. Similarly as with national renewable targets, some Member

States opt for a capacity payment policy to power plants in their territories. A capacity payment

represents a fixed price that is paid to generators to reward them for the capacity they can

produce to guarantee security of supply, rather than what they actually produce. With this the

benefits of integrating markets in Europe are not fully exploited since larger markets increase

competition within capacity markets and less of this capacity would be needed in total.

Moreover, depending on the magnitude of these capacity payments, this policy could distort

market outcomes significantly by impacting prices in the short-term and hence influencing

power plant operations nationally and internationally. The latter can lead to a welfare

redistribution effect between interconnected national markets. If a Member State implements a

capacity payment to ensure its generation adequacy, it will most likely positively influence the

generation adequacy of an adjacent Member State with no capacity remuneration, and

ultimately, the consumers in the first one will be the ones financing the generation adequacy

levels of the latter. Therefore, it is crucial that the common approach of the IEM is not

jeopardized by divergent policies for security of supply. Member States should be encouraged

to homogenize them by European institutions.

With regards to considering a change in the market design, welfare redistribution issues can also

arise between already existing generators. This issue is particularly relevant for a market

shifting to a nodal scheme, in which nodal prices can experiment more drastic changes than in a

zonal market, thus shifting economic surplus between actors, both generators and consumers.

As a means to mitigate these congestion costs, especially during the initial implementation

phases of the new market design, financial instruments such as financial transmission rights

(FTRs) could become an important mechanism to alleviate impacts on revenues and high price



risks. The holder of an FTR is allowed to receive the difference between two nodes prices, both

of them specified in the FTR conditions. TSOs would fund this financial instrument from the

congestion revenue it receives. Nonetheless, the increased variability and uncertainty in the

system due to large amounts of renewable energies also increase the uncertainty for FTR

holders about the congestion patterns in the lines. Therefore, new financial instruments need to

be developed to address the intermittency of renewables.




Exploration of the future European Electricity Market Design Bibliography


BIBLIOGRAPHY

Baldick, R., 2003. 'Variation of Distribution Factors with Loading'. IEEE Transactions on Power Systems, 18(4),

pp. 1316-1323.

Baldick, R., Dixit, K. & Oberbye, T., 2005. 'Empirical Analysis of the Variation of Distribution Factors with

Loading'. IEEE Power Engineering Society General Meeting, Volume 1, pp. 221-229.

Barroso, J., 2006. The EU and energy: looking to the future. EU Focus September 1-3.

Belgian Federal Government, 2013. Statistics Belgium. [Online]

Available at: http://statbel.fgov.be/en/statistics/figures/

[Accessed March 2015].

Beurskens, L., Hekkenberg, M. & Vethman, P., 2011. Renewable Energy Projections as Published in the National

Renewable Energy Action Plans of the European Member States. Energy research Centre of the Netherlands

(ECN) and European Environment Agency (EEA). [Online]

Available at: https://www.ecn.nl/projects/nreap/2010/home/

[Accessed April 2015].

Böckers, V., Haucap, P. J. & Heimeshoff, D. U., 2013. Cost of Non-Europe in the Single Market for Energy.

Benefits of an integrated European electricity market: the role of competition, Brussels: s.n.

Borggrefe, F. & Neuhoff, K., 2011. Balancing and Intraday Market Design: Options for Wind Integration. Smart

Power Market Project, s.l.: s.n.

Central Commission for Statistics, 2015. Centraal Bureau voor de Statistiek. [Online]

Available at: http://www.cbs.nl/en-GB/menu/home/default.htm


Central Statistical Office of Poland, 2015. Information Portal. Central Statistical Office of Poland. [Online]

Available at: http://stat.gov.pl/en/


Cohen, A. & Yoshimura, M., 1983. 'A Branch-and-Bound Algorithm for Unit Commitment'. IEEE Transactions on

Power Apparatus and Systems, PAS-102(2), pp. 444-451.

Dillon, T., Edwin, K., Kochs, H. & Taud, R., 1978. 'Integer Programming Approach to the Problem of Optimal

Unit Commitment with Probabilistic Reserve Determination'. IEEE Transactions on Power Apparatus and

Systems, PAS-97(6), pp. 2154-2166.

Directorate General of the Ministry of the Economy, Industry and the Digital Sector and of the Ministry of Finance

and Public Accounts, 2015. INSEE. National Institute of Statistics and Eonomic Studies. [Online]

Available at: http://www.insee.fr/en/default.asp




Doan, B.-T., 2012. Impact of German Renewable Energies on the Spot Prices of the French-German Electricity

Markets. Paris: s.n.

Ela, E., Milligan, M. & Kirby, B., 2011. NREL. Operating Reserves and Variable Generation. A comprehensive

review of current strategies, studies and fundamental research on the impact that increased RES generation has on

power system operating reserves. Technical Report, s.l.: s.n.

ENTSO-E, 2014. Overview of Internal Electricity Market-related project work, s.l.: s.n.

ENTSO-E, 2014. Scenario Outlook and Adequacy Forecast 2014-2030, s.l.: s.n.

ENTSO-E, 2014. Ten-Year Network Development Plan 2014, s.l.: s.n.

ENTSO-E, 2015. Hourly load values of all countries for a specific month. ENTSO-E Transparency Platform.

[Online]

Available at: https://www.entsoe.eu/data/data-portal/consumption/Pages/default.aspx

[Accessed 12 September 2015].

Eurelectric, 2010. Integrating Intermittent Renewables Sources into the EU Electricity System by 2020: Challenges

and Solutions, s.l.: s.n.

European Commission, 2013. The EU Emissions Trading System (EU ETS), s.l.: s.n.

European Environment Agency, 2011. National Renewable Energy Action Plan (NREAP) of Member States, s.l.:

s.n.

EWI, 2010. 'European RES-E Policy Analysis - Eine modellbasierte Studie über die Entwicklung der

Stromerzeugung aus Erneuerbaren Energiequellen in Europa und die Auswirkungen auf den konventionellen

Strommarkt', s.l.: s.n.

EWIS, 2010. 'European Wind Integration Study- Towards a successfull integration of wind power into European

Electricity Grids', s.l.: s.n.

Federal Institution under Public Law, 2015. Statistics Austria. The Information Manager. [Online]

Available at: http://www.statistik.at/web_de/statistiken/index.html


Federal Statistical Office, 2015. Statistisches Bundesamt. [Online]

Available at: https://www.destatis.de/EN/Homepage.html


Galiana, F. & Zhuang, F., 1988. 'Toward a More Rigorous and Practical Unit Commitment by Lagrangian

Relaxation'. IEEE Transactions on Power Systems, 3(2), pp. 763-770.

GAMS, 2014. General Algebraic Modeling System. GAMS Development Corporation.. [Online]

Available at: http://www.gams.com/

Garver, L., 1963. 'Power Generation Scheduling by Integer Programming Development of Theory'. IEEE

Transactions on Power Apparatus and Systems, Volume PAS-18, pp. 730-735.



Government Digital Service, 2015. Statistics UK. [Online]

Available at: https://www.gov.uk/government/statistics


Hakvoort, R., Harris, D., Meeuwsen, J. & Hesmondhalgh, S., 2009. A system for congestion management in the

Netherlands. Assessment of the options, Zwolle: s.n.

ISTAT, 2015. Italian National Institute of Statistics. [Online]

Available at: http://www.istat.it/en/


Karan, K. & Kazdagli, H., 2011. The development of energy markets in Europe. In: Financial Aspects in Energy: A

European Perspective. Berlin, Heidelberg: Springer, pp. 11-32, 231.

Litvinov, E., 2011. Power System and LMP Fundamentals. Business Architecture and Technology. WEM 301 ISO

New England Inc. PowerPoint slides. s.l.:s.n.

Morales-España, G., Latorre, J. M. & Ramos, A., 2013. Tight and Compact MILP Formulation for the Thermal

Unit Commitment Problem, s.l.: s.n.

Muckstadt, J. & Wilson, R., 1968. 'An Application of Mixed-Integer Programming Duality to Scheduling Thermal

Generating Systems'. IEEE Transactions on Power Apparatus and Systems, pp. 1968-1978.

Neuhoff, K. et al., 2013. Renewable electric energy integration: Quantifying the value of design of markets for

international transmission capacity. Energy Economics, Volume 40, pp. 760-772.

Neuhoff, K., Hobbs, B. F. & Newbery, D., 2011. Congestion Management in European Power Networks: Criteria

to Assess the Available Options. Smart Power Market Project, s.l.: s.n.

Ouyang, Z. & Shahidehpour, S., 1991. 'An Intelligent Dynamic Programming for Unit Commitment Application'.

IEEE Transactions on Power Apparatus and Systems, 6(3), pp. 1203-1209.

Pollitt, M., 2009. Electricity Liberalization in the EU: A Progress Report, s.l.: s.n.

Safari, A., Shayanfar, H. & Jahani, R., 2010. Optimal Unit Commitment of Power System Using Fast Messy

Genetic Algorithm. International Journal on “Technical and Physical Problems of Engineering” (IJTPE).

Published by International Organization on TPE (IOTPE), 2(2), pp. 22-27.

Saha, S. et al., 2011. updated monthly. NCEP Climate Forecast System Version 2 (CFSv2). Selected Hourly Time-

Series Products. Research Data Archive at the National Center for Atmospheric Research, Computational and

Information Systems Laboratory. [Online]

Available at: http://rda.ucar.edu/datasets/ds094.1/

[Accessed 20 October 2015].

Sandrin, P. & Merlin, A., 1983. 'A New Method for Unit Commitment at Electricité de France'. IEEE Transactions

on Power Apparatus and Systems, Volume PAS-102, pp. 1218-1255.



Shahidehpour, S. & Ouyang, Z., 1992. 'A Hybrid Artificial Neural Network/Dynamic Programming Approach to

Unit Commitment'. IEEE Transactions on Power Systems, 7(1), pp. 236-242.

Sheble, G., 1990. 'Solution of the Unit Commitment Problem by the Method of Unit Periods'. IEEE Transactions

on Power Systems, 5(1), pp. 257-260.

Smeers, Y., 2008. 'Study on the general design of electricity market mechanisms close to real time'. Study for the

Commission for Electricity and Gas Regulation (CREG), s.l.: s.n.

Snyder, W., Powell, H. & Rayburn, J., 1987. 'Dynamic Programming Approach to Unit Commitment'. IEEE

Transactions on Power Apparatus and Systems, Volume PAS-2, pp. 339-350.

State Administration of the Czech Republic, 2014. Czech Statistical Office. [Online]

Available at: https://www.czso.cz/csu/czso/about-czso


Statistics Norway, 2015. StatBank Norway. [Online]

Available at: https://statbank.ssb.no/en/statistikkbanken


Statistics Sweden, 2013. SCB. Statistics Sweden. [Online]

Available at: http://www.scb.se/en_/


The Danish Ministry of Social Affairs and the Interior, 2015. Statistics Denmark. [Online]

Available at: http://www.dst.dk/en#


Tseng, C.-L.et al., 1999. A Transmission-Constrained Unit Commitment Method in Power System Scheduling.

Decision Support Systems, Volume 24, p. 297–310.

TUDelft, 2014. Enipedia. [Online]

Available at: http://enipedia.tudelft.nl/wiki/Main_Page

Van Blijswijk, M., 2011. Congestion management in the Dutch power sector: A quantitative evaluation of policy

options. Master Thesis, s.l.: s.n.

Van Blijswijk, M., de Vries, L. & Van der Lee, G., 2015. 'Modeling transmission system development in North

West Europe with endogenous investment'. Draft article submitted for publication in a scientific journal,

September 2015.

Van den Bergh, K. & Delarue, E., 2015. Cycling of conventional power plants: Technical limits and actual costs.

Energy Conversion and Management, Volume 97, pp. 70-77.

Van den Bergh, K., Delarue, E. & D'Haeseleer, W., 2014. DC Power Flow in Unit Commitment Models.TME

Working Paper - Energy and Enviroment, s.l.: s.n.

Exploration of the future European Electricity Market Design Appendices


APPENDICES

Exploration of the future European Electricity Market Design Appendices



Exploration of the future European Electricity Market Design Appendix A – GAMS Model Code


APPENDIX A

GAMS model code

WEEKLY UNIT COMMITMENT AND TRANSMISSION CONSTRAINED ECONOMIC DISPATCH OF

THERMAL AND RES UNITS (SDUC AND TCED)

$OnText

Ainhoa Villar, based on Germán Morales SDUC model. February 24, 2015

$OffText

$OnEmpty OnMulti OffListing

* Solve the optimization problems until optimality

option Optcr = 0.001, Reslim=12000;

* Declaration of sets and indices

SETS

p hourly periods

p1(p) first hour of the day

sc wind scenarios (inactive set)

g generation units

g1(g) generation unit with up time equal to 1h

i buses

l dc lines

is(i) buses without slack bus

iac(i) busses connected to at least one ac line connection

idc(i) dc busses

isac(i) busses without slack bus and connected to at least one ac line connection

c circuit ID

ij(i,i,c) Buses connected by a transmission line and circuit ID c

ijac(i,i,c) Buses connected by an ac transmission line and circuit ID c

ij2(i,i) pairs of buses with at least one line connection

ig(i,g) location of generator g in node i

alias(p,pp)

alias(i,j)

alias(c,cc)

alias(i,ii)

alias(l,ll)

;

PARAMETERS

*Parameters related to hourly periods



pDemand(p,i) demand for every hour p in every node i [GW]

pPMaxWindGen(i,p) stochastic wind generation for every scenario every hour & every node [GW]

pSwProb(sc) probability of scenario sc [pu]

pENScost cost of energy not served [M€ per GWh]

*Parameters related to generator units

pMaxProd(g) Maximum output of generator g [GW]

pMinProd(g) Minimum output of generator g [GW]

pIniProd(g) Initial output of unit g [GW]

pIniProd1(g) Initial output of unit g above minimum output [GW]

pIniState(g) Initial state of unit g [h]

pIniUC(g) Initial commitment of unit g [0-1]

pIniUp(g)

pIniDw(g)

pRampUp(g) Ramp up of unit g [GW per h]

pRampDw(g) Ramp down of unit g [GW per h]

pMinUpT(g) Minimum upward time of unit g [h]

pMinDwT(g) Minimum downward time of unit g [h]

pFuelCost(g) Cost of fuel used by generator g [M€ per GWh]

pOMCost(g) Variable operation cost & maintenaince cost for generator g [M€ per GWh]

pMargCost(g) Marginal cost for generator g [M€ per GWh]

pFixedFuelCost(g) Fixed fuel consumption of generator g [M€ per h]

pStartUpCost(g) startup cost [M€ per GW]

pShutDownCost(g) shutdown cost [M€ per GW]

pEmissionRate(g) emission rate of generator g [tCO2 per GWh]

pCO2Cost CO2 emission cost rate [M€ per tCO2]

pFuelType(g) Fuel type identification

*Parameters related to the network

pPTDF(i,j,c,i) Power Transfer Distribution Factors Matrix for line ij circuit ID c and node i

Adc(l,i) Incidence matrix of HVDC lines

pBranchMaxPower(i,j,c) Maximum power flow for line ij

pDCMax(l) Maximum power output of HVDC line l

*Parameters for post-calculations

pExpnr Experiment number

*pLMPac(p,i,sc) locational marginal price in node i for hour p and scenario sc [M€ per GWh]

*pLMPdc(p,i,sc) locational marginal price in node i for hour p and scenario sc [M€ per GWh]

pPnet(i,p,sc) Net production (Production-Demand+ENS) of evey node i in every hour p [GW]

pPnet2(i,p,sc)

pNetInflowDC(i,p,sc) Net inflow due to DC lines in every node i for every hour p [GW]

pNetInflowDC2(i,p,sc)

pNetInflowAC(i,p,sc) Net inflow due to AC lines in every node i for every hour p [GW]

pNetInflowAC2(i,p,sc)

pNetInflow(i,p,sc) Total net inflow due to all lines (DC and AC) in every node i in every hour p [GW]



pSummary(*)

pTotal(g,p,sc)

;

* Declaration of free variables

VARIABLES

vTotalVCost total system variable cost [M€]

vPBranch(i,j,c,p,sc) power flow through branch i -> j and circuit c for every hour p & scenario sc [GW]

vPDCline(l,p,sc) power flow through dc line l for every hour p and scenario sc [GW]

;

POSITIVE VARIABLES

vProduct (g,p,sc) output of the unit g for hour p and scenario sc [GW]

vProduct1 (g,p,sc) output of the unit g > min load for hour p and scenario sc [GW]

vUpReserve(p,g,sc) operating reserve up of unit g for hour p and scenario sc [GW]

vDwReserve(p,g,sc) operating reserve down of unit g for hour p and scenario sc [GW]

vENS (p,i,sc) energy not served in node i for every hour p and every scenario sc [GW]

vIG (i,p,sc) intermittent generation in node i for every hour p and every scenario sc [GW]

BINARY VARIABLES

vCommitt (g,p) commitment of the unit g in hour p [0-1]

vStartup (g,p) startup of the unit g in hour p [0-1]

vShutdown (g,p) shutdown of the unit g in hour p [0-1]

* Declaration of Equations

EQUATIONS

eTotalVCost total system variable cost [M€]

eBalance (p,sc) load generation balance in hour p for scenario sc [GW]

eThrmMax (g,p,sc) max output of the committed unit g (with uptime and downtime greater than 1h) in

hour p and scenario sc [GW]

eThrmMax2 (g,p,sc) max output of the committed unit g (with uptime lower than 2h) in hour p and

scenario sc [GW]

eThrmMax3 (g,p,sc) max output of the committed unit g (with downtime lower than 2h) in hour p and

scenario sc [GW]

eThrmMin (g,p,sc) min output of the committed unit g in hour p and scenario sc [GW]

eTotOutput(g,p,sc) total output of committed unit g in hour p [GW]

eRampUp (g,p,sc) bound on ramp up of unit g in hour p [GW]

eRampDw (g,p,sc) bound on ramp down of unit g in hour p [GW]

eMinTUp (g,p) minimum up time (committed)

eMinTDown (g,p) minimum down time (not committed)

eUCStrShut(g,p) relation between commitment startup and shutdown decision

ePBranch(p,i,j,c,sc) Network constraints of line ij circuit c for hour p and scenario sc

ePDCnodes(p,i,sc) Balance equation for DC nodes



;

* Formulation of equations:

eTotalVCost .. vTotalVCost =e= sum[(p,i,sc), pENSCost*vENS(p,i,sc) ] +

sum[(p,g,sc), pEmissionRate(g)*pCO2Cost*vProduct(g,p,sc) ] +

sum[(p,g,sc), pMargCost(g)*vProduct(g,p,sc) ] +

sum[(p,g), pFixedFuelCost(g)*vCommitt(g,p) ] +

sum[(p,g), pStartupCost(g)*vStartup(g,p)*pMaxProd(g) ] +

sum[(p,g), pShutDownCost(g)*vShutdown(g,p) ];

*Balance Demand with Production of thermal generation and IG & ENS in the system every hour p

eBalance (p,sc) .. sum[g, vProduct(g,p,sc)] + sum[i,vIG(i,p,sc)+vENS(p,i,sc)] =e=sum[i,

pDemand(p,i)];

*Capacity Limits

eThrmMax (g,p,sc) $[not(g1(g))] .. vProduct1(g,p,sc) +vUpReserve(p,g,sc)=l= (pMaxProd(g)-

pMinProd(g))*(vCommitt(g,p) - vStartup(g,p) - vShutdown(g,p+1));

eThrmMax2 (g1(g),p,sc) .. vProduct1(g,p,sc) +vUpReserve(p,g,sc)=l= (pMaxProd(g)-

pMinProd(g))*(vCommitt(g,p) - vStartup(g,p));

eThrmMax3 (g1(g),p,sc) .. vProduct1(g,p,sc) +vUpReserve(p,g,sc)=l= (pMaxprod(g)-

pMinProd(g))*(vCommitt(g,p) - vShutdown(g,p+1));

eThrmMin (g,p,sc) .. vProduct1(g,p,sc)-vDwReserve(p,g,sc) =g= 0;

*Power Output of unit g always above technical minimum

eTotOutput(g,p,sc) .. vProduct(g,p,sc) =e= pMinProd(g)*vCommitt(g,p) + vProduct1(g,p,sc) ;

*Ramping constraints

eRampUp(g,p,sc) .. vProduct1(g,p,sc)-vProduct1(g,p-1,sc)-pIniProd1(g) $p1(p) =l=

pRampUp(g) ;

eRampDw(g,p,sc) ..-vProduct1(g,p,sc)+vProduct1(g,p-1,sc)+pIniProd1(g) $p1(p) =l=

pRampDw(g) - (pRampDw(g)-(pMinProd(g)-pMinProd(g)))*vShutDown(g,p)$p1(p);

*Relation between commitment startup and shutdown decision

eUCStrShut(g,p) .. vCommitt(g,p) - vCommitt(g,p-1) - pIniUC(g) $p1(p) =e= vStartup(g,p) -

vShutdown(g,p) ;

*Minimum Time Up and Minimum Time Down

eMinTUp(g,p) $[ord(p)>=pMinUpT(g)] .. sum[pp$(ord(pp)>=(ord(p)-pMinUpT(g)+1) and

ord(pp)<=ord(p)),vStartup (g,pp)] =l= vCommitt(g,p);

eMinTDown(g,p) $[ord(p)>=pMinDwT(g)] .. sum[pp$(ord(pp)>=(ord(p)-pMinDwT(g)+1) and

ord(pp)<=ord(p)),vShutdown(g,pp)] =l= 1 - vCommitt(g,p);

*Network constraint 1



ePBranch(p,i,j,c,sc) $[ijac(i,j,c)] ..vPBranch(i,j,c,p,sc) =E= SUM[isac, pPTDF(i,j,c,isac)*(SUM [g $

ig(isac,g),vCommitt(g,p)*pMinProd(g)+ vProduct1(g,p,sc)] - pDemand(p,isac) +vENS(p,isac,sc) +

vIG(isac,p,sc) + SUM[l , Adc(l,isac) * vPDCline(l,p,sc) ] ) ] ;

*Network constraint 2

ePDCnodes(p,i,sc) $[idc(i)] ..SUM [g $ ig(i,g),vCommitt(g,p)*pMinProd(g)+ vProduct1(g,p,sc)] -

pDemand(p,i) +vENS(p,i,sc) + vIG(i,p,sc) =E= -SUM[l , Adc(l,i) * vPDCline(l,p,sc) ] ;

************************************************************************************

**********************OBTAINING PARAMETERS*************************************

sets

$include indicesV1

;

$include paramV1

;

table pDemand2(p,i) thermal generation parameters

$include demandV1

;

table pPMaxWindEnergy(p,i) thermal generation parameters

$include RESproductionV1

;

table pThermalGen(g,*) thermal generation parameters

$include thermalgenV1

;

table pNetwork(i,j,c,*) network parameters

$include ACnetworkV1

;

table pDClines(l,*) dummy generators parameters

$include DClinesV1

;

************************************************************************************

*********************SCALING PARAMETERS****************************************

* Scaling of parameters from MW and € to GW and M€ from €/MWh to M€/GWh

pDemand(p,i) = pDemand2(p,i) * 1e-3 ;

pPMaxWindGen(i,p)= pPMaxWindEnergy(p,i) * 1e-3;

pMaxProd (g) = pThermalGen(g,'MaxProd') * 1e-3 ;

pMinProd (g) = pThermalGen(g,'MinProd') * 1e-3 ;

pIniProd (g) = pThermalGen(g,'IniProd') * 1e-3 ;

pIniState (g) = pThermalGen(g,'IniState') ;

pRampUp (g) = pThermalGen(g,'RampUp') * 1e-3 * 100 ;

pRampDw (g) = pThermalGen(g,'RampDw') * 1e-3 * 100;

pMinUpT (g) = pThermalGen(g,'MinUpTime' ) ;

pMinDwT (g) = pThermalGen(g,'MinDwTime' ) ;

pFixedFuelCost (g) = pThermalGen(g,'FixedFuelCost') * 1e-6 ;



pMargCost (g) = pThermalGen(g,'MarginalCost') * 1e-3 ;

pShutDownCost (g) = pThermalGen(g,'ShutDownCost' )*1e-3 ;

pStartUpCost (g) = pThermalGen(g,'StartUpCost' ) * 1e-3 ;

pEmissionRate(g) = pThermalGen(g,'EmissionRate' ) * 1e3 ;

pFuelType(g) = pThermalgen(g,'FuelType');

pBranchMaxPower(i,j,c) = pNetwork(i,j,c,'Pmax') * 1e-3;

pDCmax(l) =pDClines(l,'Pmax') * 1e-3;

************************************************************************************

***************LOADING NETWORK INFO, ASSIGNMENT OF SETS***********************

g1(g) $[pMinUpT(g)=1] = yes;

execute_load 'PTDF.gdx' isac,is,pPTDF,ijac, Adc, iac, idc;

display pPTDF ;

*Assignment of thermal generators g to their corresponding buses i

ig(i,g) $[ord(i) = pThermalGen(g,'Bus')] = YES;

display ig ;

is(i) = is(i)*(SUM[g,ig(i,g)]);

display is ;

ij(i,j,c) $ pNetwork(i,j,c,'Pmax') = YES;

display ij ;

************************************************************************************

***********BOUNDS ON VARIABLES, OTHER DATA & INITIAL CONDITIONS**************

*Allowing curtailment of Wind and Solar

vIG.up(i,p,sc)= pPMaxWindGen(i,p) ;

* Bounds on variables

vStartup.up(g,p) = 1 ;

vShutdown.up(g,p) = 1 ;

vStartup.lo(g,p) = 0 ;

vShutdown.lo(g,p) = 0 ;

vProduct1.up (g,p,sc) = pMaxProd(g)-pMinProd(g);

vUpReserve.up (p,g,sc) = pMaxProd(g)-pMinProd(g);

vDwReserve.up (p,g,sc) = pMaxProd(g)-pMinProd(g);

vPBranch.up(i,j,c,p,sc) = pBranchMaxPower(i,j,c);

vPBranch.lo(i,j,c,p,sc) = -pBranchMaxPower(i,j,c);

vPDCline.up(l,p,sc) = pDCmax(l);

vPDCline.lo(l,p,sc) = -pDCmax(l);

*Other entry data

* Determine the first hour of the day

p1(p) $[ord(p) = 1] = yes ;



* Definition of the cost of non-served energy M€/GWh

pENScost=0.2;

*Definition of the cost of CO2 emission cost rate M€/tonnes of CO2

pCO2Cost=0.0001;

* If the initial output of the unit is above its minimum load then the unit is committed, otherwise it

is not committed

pIniUC(g)= 1 $[pIniProd(g) >= pMinProd(g)] ;

pIniProd1(g)=(pIniProd(g)-pMinProd(g))*pIniUC(g);

display pIniProd1, pIniUC ;

* If the minimum up or down times are 0, they are changed to 1h

pMinUpT(g) $[pMinUpT(g) = 0] = 1 ;

pMinDwT(g) $[pMinDwT(g) = 0] = 1 ;

*Initial up/dw/Start-Up conditions

pIniUp(g) $[pIniState(g)>=0] = pIniState(g) ;

pIniDw(g) $[pIniState(g)< 0] = abs(pIniState(g));

* SOLVER OPTIONS

* Selection of the solver for linear programming problems

OPTION LP = Cplex;

* Maximum number of iterations that are allowed

*OPTION ITERLIM = 100000;

file COPT / cplex.opt /

put COPT putclose 'rinsheur 200' /;

model UCandTCED / all/ ;

UCandTCED.solprint = 0 ;

UCandTCED.holdfixed = 1 ;

UCandTCED.optfile = 1;

*model UCandTCED / eTotalVCost, eBalance, eOpReserve, eReserveUp, eReserveDw,

* eMaxOutput, eMinOutput, eTotOutput, eRampUp, eRampDw,

* eUCStrShut, eMinTUp, eMinTDown, ePBranch

* /;

* Solve the stochastic daily unit commitment & transmission constrained economic dispatch model

solve UCandTCED using MIP minimizing vTotalVCost;

pSummary('NumEqs') = UCandTCED.Numequ;

pSummary('NumRealVar') = UCandTCED.Numvar - UCandTCED.Numdvar - 1;

pSummary('NumBinVar') = UCandTCED.Numdvar;

pSummary('OptMILP') = UCandTCED.Objval;

pSummary('CPUTime') = UCandTCED.Resusd;

pSummary('NonZeros') = UCandTCED.Numnz;

pSummary('OptRMIP') = UCandTCED.Objval+eps;



pSummary('RGap') = (UCandTCED.Objest-UCandTCED.Objval)/UCandTCED.Objval + eps;

pSummary('Iters') = UCandTCED.Iterusd+eps;

pSummary('Nodes') = UCandTCED.Nodusd+eps;

pSummary('SolStat') = UCandTCED.solvestat;

pSummary('LBound') = UCandTCED.Objest+eps;

************************************************************************************

*********************CALCULATION OF OUTPUT VARIABLES**************************

*Calculation of LMPs, nodal prices finally done in MATLAB

*For nodes that are connected to both AC and HVDC lines:

*pLMPac(p,iac,sc) = eBalance.m(p,sc) + sum[(j,c), pPTDF(iac,j,c,iac)*ePBranch.m(p,iac,j,c,sc)];

*For nodes that are connected only to HVDC lines:

*pLMPdc(p,idc,sc) = eBalance.m(p,sc) + ePDCnodes.m(p,idc,sc);

display eBalance.m, ePBranch.m, ePDCnodes.m;

*Calculation of Commitment Planning Costs, Operation and Non-served energy Costs:

parameter pPlanningCost, pOperationCost, ensCost;

pPlanningCost = sum[(p,g), pFixedFuelCost(g)*vCommitt.l(g,p)] +

sum[(p,g), pStartupCost(g)*vStartup.l(g,p)*pMaxProd(g)] +

sum[(p,g), pShutDownCost(g)*vShutdown.l(g,p)];

ensCost = sum[(p,i,sc), pENSCost*vENS.l(p,i,sc)];

pOperationCost = vTotalVCost.l - pPlanningCost - ensCost;

*Calculation of aggregated production in every node i of generators located in that node

parameter pAggrProd(i,p,sc);

pAggrProd(i,p,sc) = sum[g $ig(i,g), vProduct.l(g,p,sc)] ;

*Calculation of Net inflow (Inflow-Outflow) in every node i for every hour p due to HVDC lines

pNetInflowDC(i,p,sc) = SUM[l , Adc(l,i) * vPDCline.l(l,p,sc) ] ;

pNetInflowDC2(i,p,sc) = pNetInflowDC(i,p,sc) + EPS;

*Calculation of Net inflow (Inflow-Outflow) in every node i for every hour p due to AC lines

pNetInflowAC(i,p,sc) = sum[(j,c) $ij(j,i,c), vPBranch.l(j,i,c,p,sc)] - sum[(j,c) $ij(i,j,c),

vPBranch.l(i,j,c,p,sc)];

pNetInflowAC2(i,p,sc) = pNetInflowAC(i,p,sc) + EPS;

*Calculation of Net inflow in every node i for every hour p due to AC and HVDC lines

pNetInflow(i,p,sc) = pNetInflowDC2(i,p,sc) + pNetInflowAC2(i,p,sc);

*Renaming pDemand to reposition indexes

parameter pDemand3(i,p);

pDemand3(i,p)= pDemand(p,i);

*Renaming pENS to reposition indexes



parameter pENS(i,p,sc);

pENS(i,p,sc)=vENS.l(p,i,sc);

*Calculation of hourly curtailment and % of hourly RES Curtailment in the whole system

parameter pTotalCurtailment(p,sc), pCurtailment(p,sc), pPMaxWindGen2(i,p);

pPMaxWindGen2(i,p)=pPMaxWindGen(i,p) + EPS;

pTotalCurtailment(p,sc)=sum[i, (pPMaxWindGen2(i,p) - vIG.l(i,p,sc)) ];

pCurtailment(p,sc)= pTotalCurtailment(p,sc) / [sum[i, pPMaxWindGen2(i,p)]] * 100;

************************************************************************************

*******************WRITE OUTPUT VARIABLES IN TEXT FILE**************************

file summaryout /"Summary.txt"/ ; summaryout.pc=6;

put summaryout ;

put "----------------" /

put "General information" /

put "----------------" /

put 'Expnr ' , pExpnr:<3:0 /

put "program execution date ", system.date /

put "program execution time", system.time /

put "title of the model", system.title/

put "input file name", system.ifile /

put "output file name", system.ofile /

*put "current file page", system.page /

*put "restart file date", system.rdate /

*put "restart file name", system.rfile /

*put "restart file time", system.rtime /

*put "save file name", system.sfile /

put "----------------" /

put "Solution Summary" /

put "----------------" /

put "NumEqs " , pSummary('NumEqs'):<8:0 /

put "NumRealVar" , pSummary('NumRealVar'):<8.0 /

put "NumBinVar " , pSummary('NumBinVar'):<8.0 /

put "OptMILP " , pSummary('OptMILP'):<8.6 /

put "CPUTime " , pSummary('CPUTime'):<8.6 /

put "NonZeros " , pSummary('NonZeros'):<8.0 /

put "OptRMIP " , pSummary('OptRMIP'):<8.6 /

*put "IntGap " , pSummary('IntGap'):<8.0 /

*put "OptRMIP " , pSummary('OptRMIP'):<8.0 /

put "RGap " , pSummary('RGap'):<8.6 /

put "Iters " , pSummary('Iters'):<8.0 /

put "Nodes " , pSummary('Nodes'):<8.0 /

put "SolStat " , pSummary('SolStat'):<8.0 /

put "LBound " , pSummary('LBound'):<8.6 /

put "----------------" / ;

*Power flows through AC and HVDC lines



file actxtout /"AClineFlows.txt"/ ; actxtout.pc=5;

put actxtout ;

put '# Exprnr' pExpnr:<3:0 /

loop((i,j,c,p,sc), put$(ijac(i,j,c)), i.tl, j.tl, c.tl, p.tl, vPBranch.l(i,j,c,p,sc):<15:8 /);

file dctxtout /"DClineFlows.txt"/ ; dctxtout.pc=5;

put dctxtout ;


loop((l,p,sc), put, l.tl, p.tl, vPDCline.l(l,p,sc):<15:8 /);

*Production of generators

file prodtxtout /"Production.txt"/ ; prodtxtout.pc=5;

put prodtxtout ;


loop((g,p,sc), put, g.tl, p.tl, vProduct.l(g,p,sc):<20:10 /);

*Generator Type

file gentypetxtout /"GeneratorType.txt"/ ; gentypetxtout.pc=5;

put gentypetxtout ;


loop((g), put, g.tl, pThermalgen(g,'FuelType'):<1:0 /);

*RES production

file restxtout /"RESproduction.txt"/ ; restxtout.pc=5;

put restxtout ;


loop((i,p,sc), put, i.tl, p.tl, vIG.l(i,p,sc):<20:10 /);

file maxrestxtout /"MaximumRES.txt"/ ; maxrestxtout.pc=5;

put maxrestxtout ;


loop((i,p), put, i.tl, p.tl, pPMaxWindGen2(i,p):<10:8 /);

*Aggregated production in every node of the generators located in that node

file aggrprodtxtout /"AggregatedProduction.txt"/ ; aggrprodtxtout.pc=5;

put aggrprodtxtout ;


loop((i,p,sc), put, i.tl, p.tl, pAggrProd(i,p,sc):<10:8 /);

*Demand

file demtxtout /"Demand.txt"/ ; demtxtout.pc=5;

put demtxtout ;


loop((i,p), put, i.tl, p.tl, pDemand3(i,p):<20:10 /);

*Total curtailment and % curtailment

file totalcurttxtout /"TotalCurtailment.txt"/ ; totalcurttxtout.pc=5;

put totalcurttxtout ;




loop((p,sc), put, p.tl, pTotalCurtailment(p,sc):<10:8 /);

file curttxtout /"Curtailment.txt"/ ; curttxtout.pc=5;

put curttxtout ;


loop((p,sc), put, p.tl, pCurtailment(p,sc):<6:3 /);

file netintxtout /"NetInflow.txt"/ ; netintxtout.pc=5;

put netintxtout ;


loop((i,p,sc), put, i.tl, p.tl, pNetInflow(i,p,sc):<15:8 /);

*Non-served energy

file enstxtout /"ENS.txt"/ ; enstxtout.pc=5;

put enstxtout ;


loop((i,p,sc), put, i.tl, p.tl, pENS(i,p,sc):<20:10 /);

*Costs

file coststxtout /"Costs.txt"/ ; coststxtout.pc=5;

put coststxtout ;

put '# Exprnr',pExpnr:<3:0 /

put 'Generation Costs',vTotalVcost.l:<20:3 /

put 'Planning Costs',pPlanningCost:<20:3 /

put 'Operation Costs',pOperationCost:<20:3 /;

*Commitment, Startup and Shutdown

file commtxtout /"Commitment.txt"/ ; commtxtout.pc=5;

put commtxtout ;


loop((g,p), put, g.tl, p.tl, vCommitt.l(g,p):<1:0 /);

file startxtout /"Startup.txt"/ ; startxtout.pc=5;

put startxtout ;


loop((g,p), put, g.tl, p.tl, vStartup.l(g,p):<1:0 /);

file shuttxtout /"Shutdown.txt"/ ; shuttxtout.pc=5;

put shuttxtout ;


loop((g,p), put, g.tl, p.tl, vShutdown.l(g,p):<1:0 /);

*Network Information

file acnetwtxtout /"ACNetwork.txt"/ ; acnetwtxtout.pc=5;

put acnetwtxtout ;


loop((i,j,c), put$(ijac(i,j,c)), i.tl, j.tl, c.tl, pNetwork(i,j,c,'Pmax'):<15:8 /);

file dcnetwtxtout /"DCNetwork.txt"/ ; dcnetwtxtout.pc=5;



put dcnetwtxtout ;


loop((l), put, l.tl, pDClines(l,'Pmax'):<15:8 /);

file ptdfout /"PTDF.txt"/ ; ptdfout.pc=5;

put ptdfout ;

put "Expnr" , pExpnr:<3:0 /

put "fromN", "toN", "c", loop(i, put$(isac(i)), i.tl) put / ;

loop((i,j,c), put$(ijac(i,j,c)), i.tl, j.tl, c.tl, loop(ii, put$(pPTDF(i,j,c,ii)), pPTDF(i,j,c,ii):<10:8 )

put$(ijac(i,j,c)) / );

*Dual Variables

file lambdaout /"Lambda.txt"/ ; lambdaout.pc=5;

put lambdaout ;


loop((p,sc), put, p.tl, eBalance.m(p,sc):<20:10 /);

file muout /"Mu.txt"/ ; muout.pc=5;

put muout ;


loop((i,j,c,p,sc), put$(ijac(i,j,c)), i.tl, j.tl, c.tl, p.tl, vPBranch.m(i,j,c,p,sc):<20:10 /);

$ontext

file mudcout /"MuDC.txt"/ ; mudcout.pc=5;

put mudcout ;


loop((l,p,sc), put, l.tl, p.tl, vPDCline.m(l,p,sc):<20:10 /);

$offtext

file lambdadcout /"LambdaDC.txt"/ ; lambdadcout.pc=5;

put lambdadcout ;


loop((i,p,sc), put$(idc(i)), i.tl, p.tl, ePDCnodes.m(p,i,sc):<20:10 /);

************************************************************************************

Exploration of the future European Electricity Market Design Appendix B- Generation mix of the system, per country and per scenario


APPENDIX B

Generation mix of the system, per country and per scenario

Table B.1. Generation mix of the system. Total capacity installed (GW) in every scenario.

Wind Solar RES Nuclear Lignite

Hard Coal

(Anthracit

e + Coal)

Gas

(Natural

Gas + Gas)

Fuel Oil

(Light +

Heavy)

Total

Scenario B

2014 76,5 59,0 135,5 104,2 37,5 88,6 130,4 34,1 530,2

Scenario EU

2020 157,2 94,6 251,8 99,5 34,3 76,2 135,4 25,0 622,1

Vision 1

2030 157,0 96,2 253,2 87,1 27,6 61,2 147,4 5,9 582,4

Vision 2

2030 157,0 96,2 253,2 85,3 27,6 61,2 136,8 5,9 570,0

Vision 3

2030 256,5 158,6 415,1 77,5 30,8 51,7 172,0 9,6 756,6

Vision 4

2030 325,5 189,1 514,6 77,9 26,4 45,6 167,8 9,6 841,9

Table B.2. Generation mix of the system. Increased installed capacity (%) in every scenario.

Wind Solar RES Nuclear Lignite

Hard Coal

(Anthracit

e + Coal)

Gas

(Natural

Gas + Gas)

Fuel Oil

(Light +

Heavy)

Total

Scenario

B 2014

Scenario

EU 2020

(1)

105,56 60,25 85,83 -4,55 -8,30 -14,00 3,81 -26,76 17,33

Vision 1

2030 (2) -0,13 1,74 0,57 -12,46 -19,55 -19,63 8,85 -76,46 -6,39

Vision 2

2030 (3) -0,13 1,74 0,57 -14,26 -19,55 -19,63 1,04 -76,47 -8,37

Vision 3

2030 (4) 63,16 67,73 64,88 -22,11 -10,36 -32,09 27,03 -61,66 21,63

Vision 4

2030 (5) 107,08 99,99 104,42 -21,62 -23,19 -40,15 23,91 -61,66 35,33

(1): with respect to B 2014

(2): with respect to EU 2020






Thermal, Wind & Solar capacity installed (%) associated to the national yearly targets for

RES shares in electricity in 2014-2030.

Source: ENTSO-E (Scenario B for 2014, Scenario EU for 2020 and the Visions for 2030).

Table B.3. Generation mix per country. Capacity installed (%) in scenario B 2014.

Country Wind Solar Nuclear Lignite

Hard

Coal Gas Fuel Oil

% % % % % % %

Netherlands 9 2 2 0 19 65 3

Germany 21 22 7 13 18 17 2

Austria 19 4 0 0 0 77 0

Belgium 10 15 33 0 2 39 1

France 9 5 63 0 5 6 12

Norway 41 0 0 0 0 59 0

Denmark 48 6 0 0 24 22 0

Sweden 20 0 50 0 1 5 24

UK 12 0 13 0 28 44 3

Poland 11 0 0 26 60 3 0

Czech

Republic 2 13 23 46 8 8 0

Italy 9 19 0 0 9 44 18

Total (%) 14 11 20 7 17 25 6

Table B.4. Generation mix per country. Capacity installed (%) in scenario EU 2020.


Hard

Coal Gas Fuel Oil

% % % % % % %

Netherlands 21 9 1 0 12 54 3

Germany 33 25 4 9 13 14 1

Austria 24 3 0 0 0 73 0

Belgium 22 18 23 0 0 36 0

France 21 12 52 0 2 6 7

Norway 68 0 0 0 0 32 0

Denmark 56 10 0 0 10 17 6

Sweden 33 0 52 0 1 4 12

UK 27 0 12 0 20 40 1

Poland 16 1 0 22 55 6 0

Czech

Republic 3 14 21 45 6 10 0

Italy 12 27 0 0 8 38 14

Total (%) 25 15 16 6 12 22 4



Table B.5. Generation mix per country. Capacity installed (%) in Vision 1, 2030.


Hard

Coal Gas Fuel Oil

% % % % % % %

Netherlands 21 14 2 0 18 45 0

Germany 33 31 0 8 16 10 1

Austria 20 5 0 0 7 65 2

Belgium 23 19 0 0 0 58 0

France 20 12 57 0 2 8 2

Norway 76 0 0 0 0 24 0

Denmark 48 8 0 0 20 24 0

Sweden 37 0 59 0 0 0 4

UK 32 2 11 0 4 50 1

Poland 27 2 14 22 29 6 0

Czech

Republic 4 10 26 29 8 23 0

Italy 14 25 0 0 12 47 2

Total (%) 27 17 15 5 11 25 1



Hard

Coal Gas Fuel Oil

% % % % % % %

Netherlands 21 14 2 0 18 45 0

Germany 34 31 0 8 16 9 1

Austria 21 5 0 0 8 64 2

Belgium 23 19 0 0 0 58 0

France 20 12 57 0 2 7 2

Norway 76 0 0 0 0 24 0

Denmark 49 8 0 0 21 23 0

Sweden 41 0 54 0 0 0 4

UK 34 2 12 0 4 47 1

Poland 27 2 14 22 29 6 0

Czech

Republic 4 11 27 31 8 20 0

Italy 14 26 0 0 12 46 3

Total (%) 28 17 15 5 11 24 1





Hard

Coal Gas Fuel Oil

% % % % % % %

Netherlands 29 19 0 0 13 39 0

Germany 37 30 0 6 10 18 1

Austria 30 19 0 0 7 43 2

Belgium 32 22 0 0 0 46 0

France 31 23 31 0 1 10 3

Norway 85 0 0 0 0 15 0

Denmark 53 17 0 0 14 16 0

Sweden 49 4 44 0 0 0 3

UK 46 5 12 0 5 32 1

Poland 25 2 15 29 10 18 0

Czech

Republic 3 9 34 26 7 21 0

Italy 16 36 0 0 8 35 4

Total (%) 34 21 10 4 7 23 1



Hard

Coal Gas Fuel Oil

% % % % % % %

Netherlands 30 22 1 0 11 36 0

Germany 45 27 0 5 7 15 0

Austria 26 31 0 0 5 36 2

Belgium 33 24 0 0 0 42 0

France 33 31 25 0 1 8 2

Norway 93 0 0 0 0 7 0

Denmark 55 16 0 0 13 16 0

Sweden 62 3 33 0 0 0 2

UK 48 5 11 0 5 31 0

Poland 34 12 13 16 9 16 0

Czech

Republic 5 15 32 24 5 19 0

Italy 17 37 0 0 7 35 4

Total (%) 38,66 22,46 9 3 5 20 1



Wind & Solar capacities installed (MW) associated to each country's yearly targets for

RES shares in electricity in 2014-2030.

Source: ENTSO-E (Scenario B for 2014, Scenario EU for 2020 and the Visions for 2030)

Table B.9. Installed capacities for wind and solar technologies in every country and in every

scenario (MW).

Country Scenario B 2014 EU 2020 Vision 1 2030

Wind Solar Wind Solar Wind Solar

Netherlands 2710 760 7780 3500 6000 4000

Germany 34720 36330 65790 49670 59300 55100

Austria 1900 400 2600 320 3290 820

Belgium 1720 2680 4900 4050 4790 4050

France 9000 5100 25000 15000 20000 12000

Norway 800 0 2800 0 2750 0

Denmark 4800 560 6850 1250 6850 1110

Sweden 3980 0 6400 0 6250 0

UK 7790 0 20310 0 30570 1870

Poland 3440 0 5850 270 8400 500

Czech Republic 310 2150 570 2500 740 2000

Italy North

(60% of installed

capacity in whole

country)

5298 11028 8340 18000 8046 14760

Total (GW) 76,468 59,008 157,19 94,56 156,986 96,21

Table B.10. Installed capacities for wind and solar technologies in every country and in every

scenario (MW) (continuation).

Country Vision 2 2030 Vision 3 2030 Vision 4 2030

Wind Solar Wind Solar Wind Solar

Netherlands 6000 4000 12000 8000 12800 9100

Germany 59300 55100 85000 68800 113100 68800

Austria 3290 820 5500 3500 5500 6500

Belgium 4790 4050 8540 5740 9370 6740

France 20000 12000 40000 30000 52400 49600

Norway 2750 0 5000 0 11400 0

Denmark 6850 1110 10460 3430 11460 3430

Sweden 6250 0 11100 1000 19000 1000

UK 30570 1870 54870 5800 60370 5800

Poland 8400 500 10000 1000 15600 5300

Czech Republic 740 2000 740 2000 1250 3500



Italy North

(60% of installed

capacity in whole

country)

8046 14760 13260 29340 13260 29340

Total (GW) 156,986 96,21 256,47 158,61 325,51 189,11

Figure B.1 . Breakdown of generation mix per country in Vision 2 year 2030

Figure B.2 . Breakdown of generation mix per country in Vision 3 year 2030

Exploration of the future European Electricity Market Design Appendix C- Demand parameter and assumptions


APPENDIX C

Demand parameter and assumptions

Table C.1. Annual percentage increase of demand (%) applied in scenario EU 2020.

Netherlands Germany Austria Belgium France Norway

Increase of the gross final

energy consumption in the

electricity sector from

2014 to 2020 (%)

5,3 -5,43 10,65 9,03 1,38 0,45

Note: For the additional energy efficiency scenario. Own calculation from the ECN database.

Denmark Sweden

Great

Britain Poland

Czech

Republic Italy

Increase of the gross final

energy consumption in the

electricity sector from

2014 to 2020 (%)

1,03 0,93 1,25 13,18 10,11 2,92

Note: For the additional energy efficiency scenario. Own calculation from the ECN database.

Table C.2. Adjustment of the demand in Norway (node i14 ).

Norway (node i14)

Capacity (MW) Fossil Fuels Wind Solar Total Capacity minus

Hydro

Total Capacity

(Thermal+RES+hydro)

Scenario B 2014 1150 800 0 1950 32950

Scenario EU 2020 1300 2800 0 4100 37100

Vision 1 2030 855 2750 0 3605 41505

Vision 2 2030 855 2750 0 3605 41505

Vision 3 2030 855 5000 0 5855 44455

Vision 4 2030 855 11400 0 12255 64255

Table C.3. Adjustment of the demand in Austria (nodes i09 and i10).

Austria (nodes i109 and i10)

Capacity (MW) Fossil Fuels Wind Solar Total Capacity minus

Hydro

Total Capacity

(Thermal + RES + hydro)

Scenario B 2014 8000 1900 400 10300 23900

Scenario EU 2020 8000 2600 320 10920 25020

Vision 1 2030 12156 3290 820 16266 35772

Vision 2 2030 11629 3290 820 15739 35244

Vision 3 2030 9500 5500 3500 18500 38606

Vision 4 2030 8856 5500 6500 20856 42592

Exploration of the future European Electricity Market Design Appendix C- Demand parameter and assumptions


Table C.4. Nodes list, latitude and longitude of nodes, provinces covered in each zone and

population in each zone.

Node

(Country - Zone) Nodes Latitude Longitude Provinces Population

Netherlands

NL - North i01 52,79 6,26 Groningen, Friesland, Drenthe, Overijssel,

Flevoland 3256276

NL - South i02 51,61 5,48 1/2Utrecht, Gelderland,1/2North Brabant, Limburg 4995834,5

NL - West i03 52,06 4,86 North Holland, South Holland, Zeeland, 1/2Utrecht,

1/2North Brabant 8527464,5

Germany

DE - North i04 53,28 9,99 Hamburg, Mecklenburg-Vorpommern, Schleswig-

Holstein, 1/2Niedersachsen 10054081,5

DE - West i05 51,81 7,96 Bremen, Nordrhein-Westfalen, 1/2Niedersachsen 22124526,5

DE - South i06 48,87 11,73 Bayern, Sachsen, Thüringen 18811469

DE - South West i07 49,84 8,53 Baden-Württemberg, Hessen, Rheinland-Pfalz,

Saarland 21661787

DE - East i08 51,97 12,40 Berlin, Brandenburg, Sachsen-Anhalt 8115599

Austria

AT - East i09 47,32 11,81 Vienna, Burgenland, Lower Austria, Styria,

1/2Carinthia,1/2Upper Austria 5879992,5

AT - West i10 47,79 15,18 Vorarlberg, Tyrol, Salzburg, 1/2Carinthia,1/2Upper

Austria 2619766,5

Belgium

BE - North i11 51,03 4,51 West-Flanders, East-Flanders, Antwerpen,

Limburg, 1/2Flemish Brabant, 1/2Brussels 6408536,5

BE - South i12 50,37 5,26 Hainaut, 1/2Flemish Brabant, Walloon Brabant,

Namur, Liege, Luxembourg, 1/2Brussels 4691017,5

France i13 48,11 3,49 65754000

Norway i14 60,58 8,30 5109044

Denmark

DK - West i15 56,19 9,20 Nordjylland, Midtjylland, Syddanmark 3071110

DK - East i16 55,52 11,77 Hovedstaden, Sjaelland 2588605

Sweden i17 57,69 14,26 9 644 864

Great Britain i18 52,16 -0,48 64100000

Poland i19 52,87 16,98 38495700

Czech Republic i20 49,71 14,15 10510719

Italy

IT - North West i21 45,35 8,66 Aosta Valley, Piedmont, Liguria, Lombardy 16130725

IT - North East i22 45,68 11,47 Trentino-Alto Adige, Veneto, Emilia-Romagna,

Friuli-Venezia Giulia 11654486

Exploration of the future European Electricity Market Design Appendix D- Generator properties, fuel costs and network parameters


APPENDIX D

Generator properties, fuel costs and network parameters

Table D.1. Fuel costs according to type of fuel (€/MWh)

Nuclear 3.0

Lignite 4.40

Coal 12.70

Anthracite 12.70

Gas 38.70

Natural Gas 38.70

Light Fuel Oil 30.60

Heavy Fuel Oil 30.60

Table D.2. Fixed Fuel Cost according to fuel type (€/h).

Fuel type

Fixed fuel

consumption

(MThermie/h)

Random cost range for

fixed fuel consumption cost

(€/h)

Nuclear 0 10 - 30

Lignite 0.015 10 - 30

Coal 0.03 10 - 30

Anthracite 0.06 10 - 30

Gas 0.09 30 - 70

Natural Gas 0.09 30 - 70

Light Fuel Oil 0.08 30 - 70

Heavy Fuel Oil 0.08 30 - 70

The Fixed Fuel Consumption (MTh/h) according to the fuel type is used to specify an indicative

cost range for the Fixed Fuel Cost (€/h). This means that it will be assumed that those fuel types

like nuclear, lignite, coal and anthracite, which have lower fixed fuel consumption rates

compared to natural gas, gas and light and heavy fuel oil, will also have a lower fixed fuel

consumption cost.

In the same way as it was assumed with the fuel cost (€/MWh), random variability is introduced

by using a cost range of fixed fuel consumption costs for each type of fuel. Table D.2 shows the

cost ranges chosen for the calculation.



Table D.3. Operation & maintenance costs according to type of fuel (€/MWh)

Nuclear 1.2

Lignite 2.4

Coal 1.8

Anthracite 1.2

Gas 2

Natural Gas 1.2

Light Fuel Oil 1.2

Heavy Fuel Oil 1.2

Table D.4. Average operating efficiency according to type of technology (%)

Nuclear 30

Lignite 36

Coal 38

Anthracite 38

Gas 39

Natural Gas 50

Light Fuel Oil 39

Heavy Fuel Oil 39

Table D.5. Average startup costs according to type of technology (€/MW)

Generation technology Direct Cost Indirect Cost Total cost

Nuclear 35 0 35

Lignite-fired steam 28 55 83

Coal-fired steam 25 55 80

Anthracite-fired steam 25 55 80

Combined-cycle gas turbines 5 40 45

Gas-fired steam 33 40 73

Light & heavy fuel oil 33 40 73



Table D.6. Capacities of generating units according to fuel type for scenario B 2014 (MW)

Country Anthracite

+ Coal Lignite

Light +

Heavy Fuel

Oil

Natural

Gas + Gas Nuclear

Netherlands 953,33 138,57 607,88 490

Germany 731,71 2124 150 569,2 1508,75

Austria 150

Belgium 160 1186

France 346,67 571,43 90,63 3321,05

Norway 71,88

Denmark 133,89 3,33 84,23

Sweden 230 243,16 153,33 3300

UK 845,45 120,53 403,78 898

Poland 637,1 858 92

Czech Republic 56,82 206,22 8 42,12 930

Italy 172,97 653,5 509,88

Table D.7. Capacities of generating units according to fuel type for scenario EU 2020 (MW)

Country Anthracite

+ Coal Lignite

Light +

Heavy Fuel

Oil

Natural

Gas + Gas Nuclear

Netherlands 770 138,5 612 490

Germany 623,17 1846 104,8 558,2 1000

Austria 150

Belgium 184,2 1012

France 193,33 390,5 117,2 3315

Norway 81,25

Denmark 68,33 48,67 80

Sweden 100 121,05 116,7 3367

UK 707,3 52,1 412,57 898

Poland 635,8 809 220

Czech Republic 48,2 210,5 10 52,2 930

Italy 188 582,5 524,9



Table D.8. Capacities of generating units according to fuel type for Vision 1, 2030 (MW)

Country Anthracite

+ Coal Lignite

Light +

Heavy Fuel

Oil

Natural

Gas + Gas Nuclear

Netherlands 869 0 396,94 484

Germany 701,8 1486,7 52,04 370,46 0

Austria 204,19

Belgium 287,26 0

France 116 83,48 116,61 2947,37

Norway 53,44

Denmark 162,5 0 134,85

Sweden 0 34,74 0 3317,33

UK 174,14 26,53 638,95 1092,1

Poland 297,13 696 195

Czech Republic 68,18 156,76 0 139,7 1300

Italy 235,34 89,74 553,58


Country Anthracite

+ Coal Lignite

Light +

Heavy Fuel

Oil

Natural

Gas + Gas Nuclear

Netherlands 869 0 394,33 484

Germany 701,81 1486,72 52,05 318,67 0

Austria 194,05

Belgium 278,98 0

France 116 83,48 107,58 2947,37

Norway 53,44

Denmark 162,5 0 123,5

Sweden 0 34,74 0 2719,67

UK 174,14 26,53 582,17 1092,1

Poland 297,13 696 195

Czech Republic 68,18 156,76 0 114,33 1300

Italy 235,34 89,72 531,14




Country Anthracite

+ Coal Lignite

Light +

Heavy Fuel

Oil

Natural

Gas + Gas Nuclear

Netherlands 869 0 485,7 0

Germany 551,98 1316,5 52,04 825,26 0

Austria 153,12

Belgium 287,26 0

France 116 178,57 195 2105,26

Norway 53,44

Denmark 148,78 0 124,31

Sweden 0 34,74 0 3317,33

UK 264,45 31,89 513,58 1391

Poland 132,9 1181,7 713

Czech Republic 68,18 156,76 0 139,7 1900

Italy 235,34 189,64 573,11


Country Anthracite

+ Coal Lignite

Light +

Heavy Fuel

Oil

Natural

Gas + Gas Nuclear

Netherlands 770,17 0 461,03 484

Germany 451,79 1316,52 52,05 786,04 0

Austria 145,39

Belgium 276,56 0

France 116 178,57 195 2105,26

Norway 53,44

Denmark 148,78 0 124,31

Sweden 0 34,74 0 3317,33

UK 264,45 31,89 513,58 1391

Poland 132,9 741 713

Czech Republic 54,55 156,76 0 136,36 1900

Italy 204,4 189,62 563,33



Table D.12. Minimum generation according to type of technology (% of nominal output)

Nuclear 40

Lignite-fired steam 40

Coal-fired steam 38

Anthracite-fired steam 38

Combined-cycle gas turbines 33

Gas-fired steam 33

Light & heavy fuel oil 33

Table D.13. Upward and downward ramp limits according to type of technology (MW/h)

Nuclear 1000


Coal-fired steam 40



Gas-fired steam 80

Light & heavy fuel oil 130

Table D.14. Minimum online and offline times according to type of technology (h)

Nuclear 40


Coal-fired steam 6



Gas-fired steam 2

Light & heavy fuel oil 0.042

Calculation of the inductive reactances of the transmission lines.

To calculate the reactances in the per unit system, an impedance base is calculated by

assuming a value for and for the whole network:

Therefore, the reference unitary reactance in p.u./km used for all lines equals to:

And using , the real reactances in the model vary between [0.51, 2.03] p.u.

I LARGE-S I RES ELECTRIC P S

Documents